Measuring State and Local
Government Labor Productivity:
Examples from Eleven Services
U.S. Department of Labor
Alexis M. Herman, Secretary
Bureau of Labor Statistics
Katharine G. Abraham, Commissioner
June 1998
Bulletin 2495
Preface
In 1984 the Bureau of Labor Statistics outlined possible procedures for preparing
State and local government labor productivity indexes in one of its bulletins.1 BLS
produced that publication in response to suggestions that it examine the feasibility of
preparing such indexes by the Joint Economic Committee of the U.S. Congress, the
National Research Council of the National Academy of Sciences, and the General Accounting Office. Following that publication, BLS conducted research on State and
local government services. This led to a number of internal reports and publication of
several measures as part of a bulletin entitled Selected Industries and Government Services, an annual BLS publication. The program was terminated in 1994 because of
budgetary constraints.
This bulletin presents some of the data and findings produced by this research. It is
presented here for the use of the State and local government research community and
others who are interested in the subject of productivity measurement. Much of the data
in this publication have not been published before; however, some of them are dated.
Furthermore, BLS revised its industry productivity calculations and procedures since
the research on State and local government was conducted. Thus, the State and local
government and private sector methodologies often diverge. These issues are noted at
several points in the text, particularly in the chapter that discusses conceptual considerations.
Donald M. Fisk and Mary M. Greiner prepared this bulletin under the supervision of
John Duke and Mary Jablonski, supervising economists, and Kent Kunze, Chief, Division of Industry Productivity Studies, in the BLS Office of Productivity and Technology. A number of individuals inside and outside BLS reviewed parts of the manuscript.
The authors would particularly like to thank Arie Halachmi, Tennessee State University; James Jarrett, University of Texas; John Neff and Larry Pham, American Public
Transit Association; and Ronald Wilus of the Employment and Training Administration, U.S. Department of Labor. Eugene Becker, Division of BLS Publishing, edited
the manuscript. Irma Mayfield, a Visual Information Specialist in the same division,
designed the bulletin.
Material in this publication is in the public domain and, with appropriate credit, may
be reproduced without permission. This material is available to sensory impaired individuals upon request. Voice phone: (202) 606-7828; TDD phone: (202) 606-5897;
TDD message referral phone: 1-800-326-2577.
1
For details, see Measuring Productivity in State and Local Government, BLS Bulletin 2166.
III
Contents
Chapters
Page
1. Introduction, summary, and conclusions .......................................................... 1
Productivity defined .................................................................................. 1
Basic measurement issues .......................................................................... 2
Services examined ..................................................................................... 5
Summary .................................................................................................... 8
Conclusions and observations .................................................................... 8
Measurement outlook ................................................................................ 13
2. Methodological considerations ........................................................................ 15
The production framework and process .................................................... 15
Measuring outputs ..................................................................................... 17
Measuring inputs ...................................................................................... 21
Other issues ................................................................................................ 26
The productivity index ............................................................................... 29
3. Enterprise services ........................................................................................... 33
Electric power .................................................................................................. 33
Institutional setting .................................................................................... 33
Outputs ....................................................................................................... 35
Labor inputs ............................................................................................... 39
Productivity indexes .................................................................................. 41
Government-private comparisons .............................................................. 41
Natural gas ....................................................................................................... 43
Institutional setting .................................................................................... 43
Outputs ....................................................................................................... 45
Labor inputs ............................................................................................... 47
Productivity indexes .................................................................................. 48
Government-private comparisons .............................................................. 48
Water supply .................................................................................................... 49
Institutional setting .................................................................................... 49
Quality concerns ........................................................................................ 51
Outputs ....................................................................................................... 51
Labor inputs ............................................................................................... 59
Productivity indexes .................................................................................. 61
Mass transit ...................................................................................................... 62
Institutional setting .................................................................................... 62
Outputs ....................................................................................................... 64
Which measure? ........................................................................................ 66
Quality and level of service ....................................................................... 67
Index construction and data availability .................................................... 68
Vehicle revenue mile indexes .................................................................... 68
Passenger trip indexes ............................................................................... 70
Weighted output indexes ............................................................................ 71
V
Contents—Continued
Page
Labor inputs ............................................................................................... 72
National productivity indexes .................................................................... 73
Modal trend comparisons ....................................................................... 74
System trend comparisons ...................................................................... 76
Modal level comparisons ....................................................................... 79
State alcoholic beverage sales ......................................................................... 80
Institutional setting .................................................................................... 80
Outputs ....................................................................................................... 83
Labor inputs ............................................................................................... 85
Productivity indexes .................................................................................. 87
Government-private comparisons .............................................................. 87
Summary and conclusions ............................................................................... 88
4. Corrections ...................................................................................................... 98
State adult correctional institutions ................................................................. 99
Institutional setting .................................................................................... 99
Outputs .......................................................................................................101
Candidate measures ................................................................................101
Inmates incarcerated: An undifferentiated output series ........................ 105
Inmates incarcerated: A security differentiated output series ................ 105
Comparison of the output indexes .......................................................... 107
Labor inputs ............................................................................................... 108
Productivity indexes .................................................................................. 110
Local jails ........................................................................................................111
Institutional setting .................................................................................... 111
Outputs .......................................................................................................113
Candidate measures ................................................................................113
Quality .................................................................................................... 115
Crowding—A special concern ............................................................... 115
Recidivism and other outcome measures ............................................... 116
Inmates confined .................................................................................... 116
Labor inputs ............................................................................................... 117
Productivity indexes .................................................................................. 119
Adult probation and parole .............................................................................. 120
Institutional setting .................................................................................... 120
Outputs .......................................................................................................123
Labor inputs ............................................................................................... 125
Conclusions ............................................................................................... 126
Juvenile correctional institutions ..................................................................... 127
Institutional setting .................................................................................... 127
Outputs .......................................................................................................129
Output measures .....................................................................................129
Quality and crowding ............................................................................. 131
Recidivism and other outcome measures ............................................... 132
The output series .................................................................................... 132
Labor inputs ............................................................................................... 134
Productivity indexes .................................................................................. 135
VI
Contents—Continued
Page
Summary and conclusions ............................................................................
State adult institutions ............................................................................
Local jails ...............................................................................................
Juvenile institutions ................................................................................
Comparison of the indexes .....................................................................
Overall averages .....................................................................................
136
137
138
138
139
139
5. State employment security services ..............................................................
Unemployment Insurance .............................................................................
Institutional setting .................................................................................
Outputs ....................................................................................................
Number of employees covered ............................................................
Number of employers covered ............................................................
Number of beneficiaries ......................................................................
Number of compensated weeks ...........................................................
Number of claims processed ...............................................................
Benefit index .......................................................................................
Finance index ......................................................................................
Program index .....................................................................................
Comparison of output indexes ............................................................
Quality of UI output ............................................................................
Labor inputs ............................................................................................
Productivity trends ..................................................................................
148
148
148
150
150
150
151
151
151
152
152
152
153
153
154
157
Employment Service .....................................................................................
Institutional setting .................................................................................
Outputs ....................................................................................................
Placements ..........................................................................................
Referrals ..............................................................................................
Services ...............................................................................................
Recommended output series ...............................................................
Labor inputs ............................................................................................
Productivity trends ..................................................................................
160
160
162
163
164
165
169
170
171
Summary and conclusions ............................................................................ 172
6. Bibliography ................................................................................................. 176
Appendix A. Comparison of Bureau of the Census classification of State
and local government functions with Standard Industrial Classification ...... 193
Tables
Introduction, summary, and conclusions:
1. Terminology of government productivity measurement .........................
3
2. Labor productivity indexes for 10 State and local government services,
1967-92 ...................................................................................................
6
3. Average annual rate of change in output, labor and labor productivity
for 10 services, selected periods .............................................................
9
4. Comparison of rates of change for output measures for three services,
selected periods 1964-92 ........................................................................ 11
5. Comparison of rate of change of weighted and unweighted output for
seven services, selected periods 1967-92 ............................................... 12
VII
Contents—Continued
Page
Methodological considerations:
6.
7.
Examples of steps in the production of selected government services ...... 17
State and local government labor compensation as a percent of total
operating expenditures for selected functions, fiscal year 1992 ................ 23
Enterprise services:
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
VIII
Percent distribution of kilowatt hours sold, customers served, and installed capacity owned by type of utility ownership, 1992 .............................. 34
Finances of State and local government electric utilities by type of government, fiscal year 1992 ........................................................................... 35
State and local government electric utility costs per kilowatt hour by
class of service, selected years ................................................................... 36
Unweighted and weighted State and local government electric utility
output indexes, 1967-92 ............................................................................ 38
Four State and local government electric utility labor indexes, 1967-92 .. 40
Indexes for State and local government electric utilities: Output, FTE
employee, total employee and output per FTE employee and output per
employee, 1967-92 .................................................................................... 42
Comparison of State and local government and investor and cooperativeowned electric utility productivity indexes, 1967-92 ................................ 43
Finances of local government natural gas utilities by type of government,
fiscal year 1992 .......................................................................................... 44
Natural gas delivered by local government utilities by class of service,
1974-92 ...................................................................................................... 45
Local government natural gas utility costs per therm by class of service,
selected years ............................................................................................. 46
Unweighted and weighted local government natural gas output indexes,
1974-92 ...................................................................................................... 47
Indexes of output, FTE employment and output per FTE for local government natural gas utilities, 1974-92 ............................................................. 48
Comparison of local government and investor-owned natural gas utility
labor productivity indexes, 1974-92 .......................................................... 49
Finances of State and local government water supply utilities by type of
government, fiscal year 1992 ..................................................................... 50
Local government water utility revenue series and indexes, 1967-92 ....... 56
Three local government water utility output indexes, 1967-92 ................. 58
Comparison of three water supply output indexes, 1965-90 ..................... 59
Total and FTE employment, local government water supply indexes,
1967-92 ...................................................................................................... 60
Local government water supply productivity indexes, 1967-92 ................ 61
Finances of State and local government mass transit by type of government, fiscal year 1992 ................................................................................ 63
Transit modes (private and public) ranked by passenger trips, vehicle
revenue miles and operating expenses, 1992 ............................................. 63
Vehicle revenue mile indexes for local government mass transit service
by mode, 1967-92 ...................................................................................... 69
Unlinked passenger trips of local government mass transit by mode,
1967-92 ...................................................................................................... 70
Weighted and unweighted output indexes for vehicle revenue miles and
unlinked passenger trips in local government mass transit service,
1967-92 ...................................................................................................... 71
Contents—Continued
Page
32. Labor indexes for local government mass transit by mode, 1967-92 ........ 73
33. Two labor productivity indexes for local government mass transit service,
1967-92 ...................................................................................................... 74
34. Productivity indexes for local government bus service, 1967-92 .............. 75
35. Productivity indexes for local government heavy rail service, 1967-92 .... 76
36. Comparison of unlinked passenger trip productivity indexes for total
sample and New York City, Chicago, Los Angeles, and Boston transit
systems, 1967-92 ....................................................................................... 77
37. Comparison of vehicle revenue mile productivity indexes for total
sample and New York City, Chicago, Los Angeles, and Boston transit
systems, 1967-92 ....................................................................................... 78
38. Type of State alcoholic beverage control distribution, 1992 ..................... 82
39. State alcoholic beverage sale indexes, 1967-92 ........................................ 85
40. State alcoholic berverage control sale productivity indexes, 1967-92 ...... 86
41. Comparison of State government and private sector alcoholic beverage
labor productivity indexes, 1972-92 .......................................................... 87
42. Enterprise service labor productivity indexes for five services, 1967-92 .. 90
43. Comparison of labor productivity trends for three government services
and private sector industries, 1967-92 ....................................................... 91
Corrections:
44. Expenditures of State and local government on corrections, fiscal year
1992 ........................................................................................................... 98
45. Percent of inmates assigned by facility security classification, selected
years ........................................................................................................... 103
46. State adult institution inmate indexes, 1973-92 ......................................... 106
47. Unweighted output indexes by security level, 1973-92 ............................. 107
48. Unit labor requirement weights by facility-security level for State government adult institutions, selected years .................................................. 107
49. Comparison of two State adult correctional output indexes, 1973-92 ....... 108
50. State government correctional institution employment by occupation,
1962, 1965, 1974, 1979, and 1984 ............................................................ 109
51. Comparison of State correctional employment and State adult correctional institutional employment indexes, 1967-92 ......................................... 110
52. State government adult correctional institution productivity indexes,
1973-92 ...................................................................................................... 111
53. Local government jail inmate index, 1970, 1978, and 1982-93 ................ 117
54. Local government jail labor index, 1970-93 ............................................. 119
55. Local government jail labor productivity indexes, 1970-93 ...................... 120
56. Number and index of probation and parole offenders under State and
local government supervision, 1976-92 ..................................................... 121
57. Definitions of probation, parole, pardon, and aftercare ............................. 122
58. Comparison of probation and parole and total corrections full-time equivalent employment, selected years ............................................................ 126
59. Types of basic juvenile facilities ................................................................ 130
60. Juvenile institutional output indexes by type of institution, 1971-93 ........ 133
61. Comparison of undifferentiated and differentiated juvenile institutional
output indexes, 1971-93 ............................................................................ 134
62. Index of full-time employees in State and local government juvenile institutions, 1971-93 ......................................................................................... 135
IX
Contents—Continued
Page
63. State and local government juvenile institution productivity indexes,
1971-93 ...................................................................................................... 136
64. Indexes of State and local government correctional institution output,
input, and productivity, 1973-92 ................................................................ 140
Employment security:
65.
66.
67.
68.
69.
70.
71.
72.
Selected Unemployment Insurance programs, 1935-92 ............................ 149
Comparison of five Unemployment Insurance output indexes, 1964-92 ..154
Number and index of State UI labor positions, 1964-92 ........................... 155
Four UI direct labor indexes, 1964-92 ...................................................... 156
Unemployment Insurance productivity index, 1964-92 ............................ 158
Comparison of four UI direct labor productivity indexes, 1964-92 .......... 159
Candidate output measures for ES services ............................................... 166
Comparison of individual ES service outputs and total ES output,
1972-87 ...................................................................................................... 169
73. Comparison of three Employment Service output indexes, 1972-87 ........170
74. Employment Service productivity index, 1972-87 .................................... 171
75. State government employment security output, labor input, and labor
producitvity indexes, 1972-87 ...................................................................173
Charts
Introduction, summary, and conclusions:
1.
Five State and local government labor productivity indexes, 1967-92 ..... 7
Methodological considerations:
2.
3.
Basic model of the production process ...................................................... 16
Sophisticated model of the production process ......................................... 16
Enterprise services:
4.
5.
6.
State and local government and private electric power utility labor productivity indexes, 1967-92 ........................................................................ 92
Local government and private natural gas labor productivity indexes,
1974-92 ...................................................................................................... 92
State government and private liquor store labor productivity indexes,
1972-92 ...................................................................................................... 92
Corrections:
7.
State government adult correctional institutions output, and labor
productivity indexes, 1973-92 .................................................................. 137
8. Local government jail output, labor, and productivity indexes, 1982-92 ..138
9. State and local government juvenile institution output, and labor
productivity indexes, 1971-92 ...................................................................139
10. Labor productivity indexes for State and local government jails,
prisons, and juvenile institutions, 1971-92 ................................................ 139
11. State and local government combined correctional institutions output,
labor and productivity indexes, 1973-92 ................................................... 140
Employment security:
12. Employment security output, labor, and labor productivity, 1972-87 .......173
X
Chapter I. Introduction, Summary,
and Conclusions
S
tate and local government employment and expenditures have risen dramatically
over the past three and one-half decades. In 1960, State and local governments
employed about 6.1 million workers, or 8.7 percent of the civilian labor force.
They also spent about $47.6 billion on the purchase of goods and services, or 9.0 percent of the gross national product. By 1995, employment had risen to 16.5 million, or
12.5 percent of the civilian labor force, and spending increased to $841.7 billion for
goods and services, or 11.6 percent of the gross national product.1
More people were employed by State and local government in 1995 than were employed in agriculture, mining, construction, transportation and public utilities, wholesale trade, or finance, insurance, and real estate. Of the major economic sectors, only
manufacturing, retail trade, and services employed more people.2
Despite the importance and growth of State and local governments, there has been
little interest in measuring the productivity of this sector.3 Over the years, particularly
in the 1970s, a number of organizations recommended research into State and local
government productivity measurement. At one time the Joint Economic Committee of
the U.S. Congress, the National Academy of Sciences, and the General Accounting
Office each suggested that consideration be given to the measurement of State and local
government productivity.4 Each group recognized the problems associated with such
an undertaking but nevertheless believed its importance warranted further investigation. But that interest waned and attention shifted to other concerns.5
Partly in response to the recommendations that emanated in the 1970s, the interest in
Congress and the concern of the Office of Management and Budget, the Bureau of
Labor Statistics (BLS) established a program to measure Federal Government productivity in the 1970s.6 More recently, BLS supported research into the feasibility of
producing State and local government productivity indexes, and for a while published
several labor productivity indexes on a continuing basis.7 Both the Federal and the
State and local government programs were terminated in the 1990s. As part of the
closure of this activity this bulletin presents some of the data and conclusions that have
been gleaned over the past several years concerning State and local government productivity measurement.
Productivity defined
Considerable confusion surrounds the discussion of the basic concepts and procedures used in government productivity measurement. Public sector productivity literature has variously defined productivity as efficiency, effectiveness, cost reduction, reinvention, total quality management, management improvement, performance measurement, methods improvement, re-engineering, work measurement, and program evaluation.8 In recent years, greater agreement seems to have emerged on the use of key
terms. Nevertheless, it is important to define the terms used here because of past and, in
some cases, continuing confusion in this area. Definitions of the more important terms
used in this bulletin are presented in table 1.
Two terms, in particular, productivity and output, raise definitional problems. This
bulletin uses them in their broad, generic sense, but they are also used to describe very
specific, technical situations. Productivity is used to describe a variety of activities,
terms, and procedures. But it also is used in its narrow economic sense to describe the
efficiency with which resources are used to produce outputs, for example, productivity
Introduction, Summary, and Conclusions—1
indexes. The term output can also create confusion. Where possible its use is restricted
to what organizations produce, that is, the services and products that are produced and
delivered to citizens. But, at times, lacking any other suitable term, it is used to describe other organizational “outputs” including activities, outcomes, and impacts.
Another term that creates confusion is “public,” like public utilities. Public utility
service can be provided by government, non-profit or private regulated companies.
Private and government organizations supply electricity, gas, water, mass transit, and
alcoholic beverage store sales. The focus of this bulletin is on State and local government provided service.
This bulletin assigns government productivity measures to one of three categories
based on the type of output measure. Output is used here in its generic or general sense.
The three categories are: 1) Measures focusing on operational issues; 2) those focusing
on organizational or program outputs, that is gross or direct outputs; and 3) those concerning organization or program consequences.
Operational measures are concerned with the internal workings or efficiency of an
organization. Work measurement, which deals with resource requirements under a given
technology or set of conditions, is a common operational measure. Intermediate activities or throughputs, such as the number of reports produced, number of audits completed, or the number of samples tested, and utilization measures, such as equipment
downtime, are other types of operational measures. Each is important for day-to-day
management of government.
The second category of productivity measures, direct outputs, is the final organizational output divided by the resources used to produce the output. The direct output
productivity measure is the one most commonly used to compute private sector productivity, and the one used in this bulletin for public sector measurement. Public sector
examples of such measures are the “tons of solid waste collected per employee hour”
for sanitation services and the “revenue gallons of water sold per employee hour” for
water utilities. These measures are also known as technical efficiency measures. They
do not address the issue of whether the service should be produced or relate them to
some desired goal. Rather, they are simply concerned with production efficiency.
The third category, consequences, addresses the issue of a program’s affect on society and whether it makes optimum use of resources to achieve its goals. This type of
measure is alternatively known as outcome, impact, effectiveness, and economic efficiency. Examples of these types of measures are “deaths prevented per employee hour”
for fire departments and “jobs created per employee hour” for economic development
agencies. Measures such as these focus on consumers and consumption whereas operational and direct output productivity measures are concerned with production relationships.
Although each of these general types of productivity measures is important, the most
common type, at least nationally, is the direct output or technical efficiency measure.
This type of measure is the one most often computed for the private sector and the one
with which this study is primarily concerned.
The measurement of government and private sector productivity is similar in many
respects. Both types of organizations produce goods and services, both compete in the
marketplace to purchase resources, and both use varying combinations of resources to
produce a product or service. As long as productivity measurements are restricted to
direct outputs and the focus is on technical production issues, there should be great
similarities in the measurement of the productivity of the two sectors.
Basic measurement
issues
Specification and measurement of output is the most difficult problem in measuring
the productivity of State and local government. The basic measure of output should be
a homogeneous physical unit, with the unit measure of output related to the resources
spent in its production. Where a government provides a single service—solid waste
disposal and drinking water are examples—the output can be simply a count of the units
Introduction, Summary, and Conclusion— 2
Table 1. Terminology of government productivity measurement
Term
Definition
Activity ......................................
A task performed by an organization to produce a desired output. Examples include
miles driven, trucks serviced, and meters read. Sometimes described as workload.
Compensation ..........................
Total labor costs including salaries, wages, and benefits.
Consequence ...........................
The desired results of government programs or services such as improved citizen safety,
increased longevity and reduced infant mortality. Often described as impacts and outcomes.
Effectiveness ............................
The degree or extent to which program goals are met, such as the percent of population
served or percent of clients successfully treated.
Efficiency ..................................
The ratio of output to inputs such as work performed per staff hour or downtime as a
percent of total hours. Includes productivity, unit costs, and technical efficiency.
Function ...................................
A government service such as police, fire, and water supply. Function and service are
used synonymously.
Goal .........................................
A statement which describes what is to be accomplished by a program, service, or
agency. A goal of public safety is to insure a safe and secure environment.
Impact ......................................
The long-term effect of a program on a community or its citizens. Impact and consequence are used synonymously. See Consequence.
Input .........................................
The resource used by an agency to produce a function, service, program, or activity.
Examples of inputs are labor, facilities, equipment, and materials.
Outcome ..................................
Short-term impact or consequence of government action or output, such as employment service job placements. See Consequence.
Output ......................................
Generically, any government operation, activity, or consequence. Specifically, the goods
or services produced by an agency. Examples of specific outputs are the gallons of
revenue water delivered, the tons of trash collected, and the kilowatt hours of electricity
sold.
Productivity ..............................
Generically, any process which improves the efficiency or effectiveness of government
services. Specifically, the efficiency with which resources are used to produce outputs;
also, technical efficiency. Examples of technical efficiency are gallons of revenue water
delivered per labor hour and tons of trash collected per labor hour.
Productivity ..............................
index
The ratio of output to input referenced to a base year.
Public .......................................
Service provided to the community or general public by the government, non-profit, or
private, regulated company.
Service .....................................
A government function such as police, fire, or education. The terms service and function
are used synonymously.
Throughput ...............................
Activities or tasks performed to produce an output. See Activity.
Unit cost ...................................
The cost of producing one unit of output.
Unit labor ..................................
cost
The labor cost (compensation) of producing one unit of output.
Unit labor ..................................
requirement
The labor required to produce one unit of output.
Workload ..................................
The amount of work performed, usually an intermediate output, such as the number of
miles driven or the number of machines serviced. See Activity.
Introduction, Summary, and Conclusions—3
of service. However, many governments produce a number of heterogeneous services,
such as fire, police, ambulance, health, and education, and it is not always easy to identify even the basic services in such cases.
Furthermore, many government services consist of a number of different subservices
or products. The Unemployment Insurance program, for example, screens applicants,
establishes eligibility, writes checks for those eligible, and audits businesses to ensure
that they contribute the required taxes. For some government services it is difficult to
even identify the outputs. For example, what are the outputs (not consequences) of
police and education services?
In addition, productivity measurement requires that the service units be homogeneous through time. However, in many instances, the scope and dimensions of government services are constantly changing. Many transit systems now provide dial-a-ride or
demand response services in addition to regularly scheduled bus service, and some
jurisdictions have added the testing of automobile emissions to safety inspections. In
both cases the service unit has changed or expanded.
Quality and level of service considerations are also important for productivity measurement because of their potential impact on the resources required to produce a unit
of service. Outputs or inputs should be adjusted when service or product shifts affect
unit costs. Movement of solid waste collection from backdoor to curbside pickup and
improvement of drinking water quality to conform with environmental standards affect
unit costs, and productivity measurement needs to account for such changes.
Selection of the proper measure of output requires a service-by-service and productby-product examination. By dividing a service, it is usually possible to identify homogeneous outputs with reasonably stable unit costs. The difficulties with this approach
are the lack of research to identify the correct units and the lack of data with which to
make the calculations.
Data to calculate State and local government output indexes often are lacking. The
Federal Government collects some data, particularly in those areas where it has shared
responsibilities, such as unemployment insurance and drinking water. National associations and public interest groups collect other data. These statistics are often inaccurate and incomplete. But more often than not, national statistics are simply unavailable
on State and local government output.
The output indexes calculated and presented in this bulletin use a base year weighted
index (Laspeyres index) when combining multiple outputs. The weights were revised
every 5 years, and the indexes were linked. BLS used this procedure in its industry
measurements for many years. Recently the Bureau shifted to Tornqvist weighting in
calculating its private industry measurements.9 For most government services, either
procedure is appropriate, and the trends should not be greatly affected by the procedure
used.
The measurement of input is often used to characterize the type of productivity measure. A common division is single factor, such as labor, or multifactor, such as labor
and capital. In reality, there is a continuum of inputs or factors of production.
The most frequently used measure of input is labor, constituting over half of all State
and local government operating expenditures. Labor is important for public policy
considerations, is easy to calculate compared with other factors of production,and is the
most accessible of State and local government factor inputs. It is the measure used
here.
The labor measure that is used most frequently for private sector measurements is
labor hours. However, no national statistics exist on State and local government labor
hours; few governments even collect such data. The measure most often used by State
and local governments is the number of full-time-equivalent employees, which is equivalent to the hours measure for calculating trends. Many governments also keep statistics
on the number of employees, a measure often used in the private sector productivity
calculations.
None of the sources of national statistics are entirely satisfactory for computing State
and local government labor productivity indexes. National statistics collected through
Introduction, Summary, and Conclusion— 4
the Bureau of the Census’ Census of Governments program, the Employment Service’s
ES-202 reports, and the Bureau of Labor Statistics’ Current Employment Statistics survey (CES-790) are not separated and categorized in small enough units to be used to
compute labor productivity indexes for individual services. In a few instances, Federal
Government programs, such as the Department of Labor’s Unemployment Insurance
program and the Department of Transportation’s Federal Transit Administration, collect State and local government employment statistics in their areas of concern. However, comparisons of labor data drawn from these and other sources reveal considerable
discrepancies from source to source.
In summary, no single source of labor data on State and local government is adequate to compute national labor indexes. Construction of viable labor indexes, like
viable output indexes, requires detailed comparison and adjustment of data.
Each productivity or labor productivity index presented in this bulletin is an output
index divided by the labor index. The term productivity, in this case, is used as an
abbreviation for output per employee. Only those services produced by State and/or
local government employees are included in the measures presented here. The period
covered by the indexes is primarily a function of data availability.
Many factors affect the indexes presented in this bulletin. Oil embargoes affect the
use of mass transit, determinant sentencing laws affect the number and type of persons
held in prison, deregulation affects the consumption of natural gas, and recessions affect the output of State unemployment insurance offices. Although such considerations
shape the production of government services, no attempt is made to assess their impact
or adjust the labor productivity indexes.
This bulletin presents labor productivity indexes as the relationship of the output of
the service to the labor required to produce the output. The indexes do not measure the
specific contribution of labor, capital, or any other factor of production. Rather, they
reflect the joint effect of many influences, including changes in technology, capital
investment, capacity utilization, skill, and effort of the work force, managerial ability,
and legislation and regulation.
Services examined
This bulletin examines 11 State and local government services and presents labor
productivity indexes for 10 of them. The more important government services, including education, police, and fire are not included because of conceptual and/or data problems.
Five of the services operate in a manner similar to private enterprises, that is, they
cover a significant portion of their expenses through fees and charges. (They will be
referred in this bulletin as “enterprise services.”) They are electric power, natural gas,
drinking water, mass transit, and alcoholic beverage sales. Four services cover the
corrections field. Calculations are presented for three—prisons, jails, and juvenile institutions. Data are not available for the fourth—probation and parole. The final two
services—unemployment insurance and the employment service—relate to employment security. Each of the 11 services is briefly discussed below. Productivity indexes
are presented for 10 services in table 2, and chart 1 shows 5 productivity indexes for 4
services. Mass transit has two different indexes. Chapter 3 discusses this variation.
Electric power. Considerable research has been conducted into private electric power
productivity, and considerable data are collected on private and public utilities. Most
electricity in the United States is generated and sold by private utilities. In 1992, however, there were about 2,000 State and local government electric power utilities, employing about 85,000 workers. From 1967 to 1992, labor productivity of State and
local government electric power systems increased at an average annual rate of 2.1
percent, output (kilowatt hours sold) was up 3.6 percent, and labor input increased 1.5
percent. Labor productivity growth was fairly robust from 1967 to 1977 but grew little
from 1977 to 1992.
Introduction, Summary, and Conclusions—5
Table 2. Labor productivity indexes for 10 State and local government services, 1967-92
(1987 = 100)
Year
Electric
power
Natural
gas
Water
supply
Mass
transit
trips1
Mass
transit
miles2
Alcohol
beverage
sales
State
prisons
1967 ............
1968 ............
1969 ............
65.0
71.7
81.5
80.5
82.3
80.6
164.9
162.1
159.1
120.2
119.8
119.5
75.7
76.2
77.1
1970 ............
1971 ............
1972 ............
1973 ............
1974 ............
1975 ............
1976 ............
1977 ............
1978 ............
1979 ............
84.9
88.0
89.5
88.8
88.1
89.6
94.5
103.4
101.5
103.7
127.4
123.6
120.9
118.8
120.1
120.0
80.1
80.4
82.2
82.4
79.9
82.1
81.2
81.1
77.8
86.7
151.6
139.9
132.8
127.5
122.4
120.7
117.2
117.9
118.0
125.3
121.5
118.4
116.4
117.3
114.1
113.4
113.7
113.9
112.2
111.4
76.4
81.0
83.9
89.2
90.0
92.9
93.6
96.7
98.6
100.4
99.1
100.6
108.0
109.4
108.3
109.5
106.3
1980 ............
1981 ............
1982 ............
1983 ............
1984 ............
1985 ............
1986 ............
1987 ............
1988 ............
1989 ............
105.2
103.2
100.7
98.7
97.3
97.8
97.4
100.0
105.0
108.1
127.7
125.5
115.8
114.0
119.1
104.7
94.4
100.0
99.0
97.0
86.1
87.4
84.3
86.3
90.5
95.8
98.3
100.0
99.4
101.4
122.6
119.5
113.0
110.8
113.1
106.6
103.4
100.0
99.3
98.8
108.0
107.9
106.0
104.6
104.1
100.6
101.4
100.0
103.0
104.6
99.8
99.7
102.0
104.0
104.9
99.2
96.4
100.0
97.9
97.8
1990 ............
1991 ............
1992 ............
110.2
110.9
108.1
92.6
95.3
95.7
99.2
93.2
92.9
97.1
93.9
97.4
106.4
105.5
106.9
98.2
98.0
100.5
Average
annual rate
of change:
1967-92 .......
1970-92 .......
1974-92 .......
1972-87 .......
2.1
1.1
1.1
.7
0.6
.7
.8
1.3
-2.1
-2.0
-1.3
-1.9
-0.5
-.6
-.4
-1.0
1.1
1.3
.6
1.2
1967-77 .......
1977-92 .......
4.7
.3
-1.4
.1
.9
-3.3
-1.3
-.5
-.4
2.5
.3
1967-72 .......
1972-77 .......
1977-82 .......
1982-87 .......
1987-92 .......
6.6
2.9
-.5
-.1
1.6
-.5
-2.9
-.9
.4
-.3
.8
3.5
-1.5
-4.2
-2.3
-.8
-2.4
-.5
-.6
-.4
-1.4
-1.2
1.3
2.1
2.9
1.1
-.4
.1
1
2
-1.6
Local
jails
Juvenile Unemploy- Employment
ment
institutions
insurance service
88.4
83.9
81.8
150.2
97.0
95.1
125.6
115.4
105.0
102.3
102.1
96.2
90.6
88.9
87.3
90.6
103.0
89.8
88.8
101.1
123.6
109.0
95.3
84.5
79.5
109.3
109.4
111.2
108.7
102.9
102.3
102.5
100.0
98.5
101.3
92.6
98.5
107.3
107.0
102.9
103.8
104.6
100.0
106.5
105.9
89.7
92.1
94.5
96.9
96.0
95.0
97.6
100.0
100.6
101.2
100.3
92.7
106.4
110.1
93.8
97.4
101.8
100.0
98.1
98.4
100.9
100.6
100.3
95.2
89.6
87.2
99.9
98.7
100.1
108.5
117.3
121.8
-.1
-.9
1.3
1.4
1.0
.7
-.5
.7
.8
1.6
.5
-2.1
.1
-4.7
.9
1.1
0
.3
1.2
2.2
-1.2
4.0
-2.4
0
-1.4
-2.7
86.6
100.4
102.4
100.5
105.5
109.4
111.9
114.1
113.1
109.1
106.8
104.7
106.1
105.4
104.7
100.0
1.0
4.8
-.5
-1.3
Unlinked passenger trips.
Vehicle revenue miles.
Natural gas. For the most part, it is the private sector that distributes and sells natural
gas. In 1992, however, about 950 local governments sold natural gas. They employed
about 11,000 individuals to handle natural gas operations. Over the 1974-92 period,
local government natural gas labor productivity decreased at an average annual rate of
1.6 percent. Output (BTU’s sold) decreased annually at 0.7 percent while labor increased 0.9 percent. There are two distinct productivity trends in the measured period:
decreases from 1974 until the mid-1980’s, then little change.
Water supply. In contrast to electric power and natural gas, drinking water operations
are largely government owned and operated. There were about 17,800 government
water systems and they employed about 157,000 workers in 1992. From 1967-92,
water-supply labor productivity increased 0.6 percent annually. Until the late 1970s
productivity growth was flat; between 1978 and 1990 it increased at a moderate rate,
but it dropped in 1991 and 1992. For the entire 1967-92 period, output, which is measured by deflated revenue, increased 1.8 percent annually while labor increased 1.2
percent.
Introduction, Summary, and Conclusion— 6
Mass transit. Both labor and output increased rapidly in the mass transit field, particularly in the 1970s and early 1980s, as government absorbed failing transit systems and
expanded service. Between 1967 and 1992, the average annual increase in labor was
2.6 percent. Two outputs, vehicle revenue miles and passenger trips, are commonly
used to assess transit output as discussed in chapter 3. The average annual increase in
vehicle miles was 2.2 percent between 1967 and 1992 and for passenger trips, it was 0.5
percent. For 1967-92, vehicle revenue mile labor productivity dropped 0.5 percent per
year while passenger trip labor productivity decreased 2.1 percent.
Alcoholic beverage sales. Seventeen States sell alcoholic beverages. Two distinct
periods mark State operations. From 1967 to 1979, there was substantial growth in
output (gallons sold) with an average annual increase of 2.9 percent. But from 1979 to
1992 output shrank at an annual rate of 2.8 percent as the States turned their operations
over to the private sector and per capita purchases of spirits dropped. The average
annual change in labor input was 0.5 percent between 1967-79 and -2.8 percent between 1979-92. Productivity grew by 2.4 percent between 1967-79 but had zero (0)
growth over the 1979-92 period. The long-term average annual change was 1.1 percent
for productivity, -0.1 for output, and -1.2 for labor input.
State prisons. One of the fastest growing State government operations is prisons. In
1990, the States operated over 1,200 prisons and employed about 246,000 persons.
Between 1974 and 1992, the average annual increase in output (number of inmates
differentiated by level of security) and labor input was the same, 7.8 percent. Average
labor productivity was flat.
Local jails. Jail output and labor input also increased in a dramatic fashion. Between
1970 and 1992, output increased at an average annual rate of 4.7 percent and labor
input increased 7.3 percent. The result was an average annual decrease in labor productivity of 2.4 percent. There were two distinct periods of change. Between 1970 and
1978, productivity decreased 5.3 percent per year, and between 1978 and 1992 the
decrease was 0.8 percent per year.
Probation and parole. The number of individuals under community probation and
parole has increased dramatically over the past several decades. In 1976, the first year
for which comparable data are available, approximately 1.1 million probation and parole offenders were under supervision; by 1992 this figure had grown to 3.4 million.
Data are lacking on individual services provided so an output index cannot be calculated. The number of employees doubled during this period.
Chart 1. Five State and local government labor productivity indexes, 1967-92
Index, 1967=100
180
180
160
160
140
140
120
120
100
100
80
80
60
60
40
Alcohol
Transit (VRM)
Electric
20
0
1967
1972
1977
40
Water
Transit (UPT)
20
1982
1987
0
1992
Introduction, Summary, and Conclusions—7
Juvenile institutions. By 1992 there were over 1,000 State and local government operated juvenile institutions, an increase of about 50 percent since 1971. However, there
were only 58,000 residents and 62,000 employees in this very labor intensive service in
1991. From 1971 to 1992 output (number of residents) was reasonably flat while labor
input rose at a fairly constant rate (1.4 percent annually). The result was an average
annual drop in labor productivity of 1.1 percent. However, there were two very distinct
periods of change during this period, which are driven by changes in output. From
1971 to 1979 output decreased at an average annual rate of 3.1 percent, but from 1979
to 1992 it increased at a 2.5 percent rate. The result of this shift is a drop of productivity (4.5 percent annually) between 1971 and 1979 and a slight increase (1.1 percent
annually) between 1979 and 1992.
Unemployment insurance. This joint Federal-State program is driven by changes in
output which largely reflect changes in the number unemployed nationally. All States
operate this program, and in 1992, it had over 46,000 State employee positions or fulltime equivalent employees. The long-term (1967-92) average annual increase in output
is 3.7 percent, for labor input it is 2.3 percent, and for labor productivity it is 1.3 percent. More interesting, perhaps, than the long-term increase is the cyclical change.
During the five cyclical periods between 1967-92, increases and decreases in output
were accompanied by increases and decreases in labor input, a situation common in the
private sector but unusual in government.
Employment service. This joint Federal-State program has been buffeted by change in
recent years, and probably has fewer than 20,000 employees today. Labor productivity
shows a 1.0 average annual percent increase from 1972 to 1987. During this period
output (services provided) remained relatively stable while labor input decreased by
0.9 percent annually. However, there were two distinct periods of change. From 1974
to 1979, labor productivity increased 2.2 percent per year, but from 1979 to 1987, it
decreased 1.6 percent annually.
Summary
Long-term labor productivity trends for the 10 measured services show productivity
increasing for 6 services and decreasing for 4. The time period covered, however, can
dramatically affect the trends of each. Only two services, natural gas and mass transit
(unlinked passenger trips), show robust and unequivocal change over the entire measured period, and both show decreasing productivity. Output and labor input increased
for most of the services as expected, given the large growth of government over the past
25 years. Two services, natural gas and alcoholic beverage sales, register long-term
decline in outputs and two, the Employment Service and alcoholic beverage sales, had
decreases in labor input (table 3).
Conclusions and
observations
Although the 10 services do not comprise a sample from which one can assess the
condition of State and local government productivity, they do provide a base for discussing some of the commonly asked questions about the subject. Also, examination of
these services should help develop additional insights into the efficiency of State and
local government operations, and possibly service productivity as a whole.
How is State and local government productivity changing? With the growth of State
and local government over the past three decades, continuing attempts to trim government operations, and the slowdown in private sector productivity, questions have frequently been raised concerning the change in government productivity. The U.S. national income accounts assume zero change in government productivity; other countries make different assumptions. Several researchers have examined and discussed the
potential impact of changing government productivity on aggregate national labor productivity statistics.10 But the discussion remains largely anecdotal and illustrative.
Clearly, the sample of 10 services selected from 3 general areas is not adequate to
answer this question, but the sample does provide additional insights to our state of
knowledge.
Introduction, Summary, and Conclusion— 8
Table 3. Average annual rate of change in output, labor, and labor productivity
for 10 services, selected periods
(in percent)
Service
(years measured)
Output
Labor
input
Labor
productivity
Electric power ...............................
(1967-92)
3.6
1.5
2.1
Natural gas ...................................
(1974-92)
-.7
.9
-1.6
Water supply .................................
(1967-92)
1.8
1.2
.6
Mass transit(upt) ...........................
(1967-92)
.5
2.6
-2.1
Mass transit(vrm) ..........................
(1967-92)
2.2
2.6
-.5
Alcohol beverage sales ................
(1967-92)
-.1
-1.2
1.1
State prisons .................................
(1973-92)
7.8
7.8
.1
Local jails ......................................
(1970-92)
4.7
7.3
-2.4
Juvenile institutions ......................
(1971-92)
.3
1.4
-1.1
Unemployment insurance ............
(1967-92)
3.7
2.3
1.3
Employment service ....................
(1972-87)
0
-.9
1.0
The sample results suggest that the long-term labor productivity trends for individual State and local government appear to be quite variable: some services increase,
some decrease, but most lack direction.
How does State and local government productivity compare with private sector productivity? The search for greater efficiency in government services often results in
calls for privatization or contracting of government operations, the argument being that
the private sector is more efficient than government. But data are rarely available with
which to test this proposition and any meaningful comparison of government and private sector productivity is notoriously difficult. The statistics presented here offer very
limited information on this subject.
In the overall aggregate business sector, there is an average annual increase in labor
productivity of 1.4 percent between 1967 and 1992. No comparable statistic exists for
State and local government.
On the other hand, many State and local government services have private sector
counterparts. This bulletin looks at three of them: Electric power, natural gas, and alcoholic beverage sales. In each case the trends of government and private sector labor
productivity are compared.
Electric power and natural gas show remarkably similar movements to their private
sector counterparts. The average annual increase in government electric power labor
productivity is 2.1 percent; for the private sector it is 2.3 percent. Both series cover
Introduction, Summary, and Conclusions—9
1967-92. For natural gas, government labor productivity dropped 1.6 percent annually
while the private sector dropped 2.2 percent; both of these series cover 1974-92. The
outputs of both the private sector and government are measured in the same way.
State alcoholic beverage sales present a slightly different picture. Government and
private sector labor productivity each increased at the same annual rate—0.9 percent—
between 1972 and 1992 but they show very different periods of growth. State government productivity increased substantially prior to 1980 while the private sector showed
its greatest growth since the mid-1980s. Some of the variances may be due to measurement differences (in the case of outputs, physical quantity was used for government and
deflated value for the private sector). Other differences may be due to different coverage (government output reflects wholesale and retail sales of alcoholic beverages only,
the private output reflects only retail sales but also includes sales of merchandise in
addition to alcohol). And some differences may be due to data problems, and some may
be due to real productivity differences.
Any comparison of private and government services, no matter how closely analyzed, is subject to numerous caveats. First, the two sectors rarely produce exactly the
same service as demonstrated by alcoholic beverage sales. Even electric power has a
different mix in the type of sales and generation. Second, rarely do comparable institutional and environmental situations exist. Third, the data are almost always drawn from
different sources and cover different periods. Finally, these comparisons of private and
government productivity are comparisons of productivity trends. Productivity levels,
such as units produced per employee, may be different.
A definitive statement about the relative efficiency of State and local government
versus the private sector based on the services examined here cannot be made. Nevertheless, it is interesting that the government and private sector productivity trends for
the three services move in the same direction, and two at about the same rate.
What is the relationship between the change in output and productivity? Analyses of
private sector operations show that large increases in output usually accompany jumps
in productivity and decreases usually lead to dropping productivity. Similar phenomena have been noted in a study of Swedish government productivity.11 Some of the
same phenomena are evident in the 10 services examined here (table 3).
With the exception of alcoholic beverage sales, a drop in output almost always resulted in a drop in labor productivity for the 10 services. Productivity for 4 of the 10
services—electric power, water, juvenile institutions, and the Employment Service—
decreased every year that output dropped. In the case of natural gas it dropped in 9 out
of 10 years and transit (unlinked passenger trips) dropped in 10 of 11 years. Employment is not adjusted in the face of falling output, at least not immediately, for these
services. Alcoholic beverage sales are the exception. In this case there were major
shifts from government to private retail operations that resulted in the closing of a large
number of State retail stores. Although service outputs decreased there were even greater
decreases in inputs with the result being increases in labor productivity.
The situation is very different when output increases. Private sector studies note that
increases in output almost always lead to increasing productivity. However, only 2 of
the 10 services, UI and alcoholic beverage sales, show a close correlation between
increasing output and increasing productivity. For the other services, output and productivity increased in the same year about half of the time. And for two services, transit
(unlinked passenger trips) and jails, increasing output was most often accompanied by
decreasing productivity. For transit trips, output and productivity increased in the same
year only 30 percent of the time, and for jails it was 27 percent of the time.
These movements are a reflection of a number of institutional factors that operate
almost independently of each other. In the case of jails, output increased rapidly throughout the entire measured period, as did staffing. However, labor input actually expanded
more rapidly than output. Employees are hired before inmates are placed in the facilities because of the need for training and employee security clearances, and the courts
have required additional staffing for many facilities. Mass transit (unlinked passenger
Introduction, Summary, and Conclusion— 10
trips), on the other hand, is a service that is supply driven. More and more service has
been provided which has resulted in greater use, but not at the same rate as employment
has increased. During the time of the oil embargoes transit output jumped ahead of
inputs but this was atypical.
Does the selection of the output measure affect the productivity trend? The measurements in this section focus on government outputs. Other measures focus on activities,
outcomes, and effects. The question becomes: Does the type of measure affect the rate
of change? The answer is that it often makes a difference, and sometimes a sizable
difference.
There are three services—transit, unemployment insurance, and employment services—for which more than one output measure was calculated. Two measures were
calculated for transit, vehicle revenue miles (VRM) which increased at an average annual rate of 2.2 percent and unlinked passenger trips (UPT) which increased at an average annual rate of 0.5 percent. Vehicle revenue miles are categorized as the service
supplied or capacity provided and the unlinked passenger trips as the output. However,
the transit industry views the revenue miles as an output and the passenger trips as an
outcome. There is considerable difference in the rate of change between the two measures, and these differences are apparent in modal calculations too.
Two measures were also calculated for Unemployment Insurance. The program
(multiple service) measure includes both benefit and tax operations; it increased 2.5
percent annually. A unitary output measure, weeks compensated, increased 3.7 percent
annually. Three measures were calculated for Employment Service. They are: Placements which is a measure of outcome; referrals which is a unitary measure of output;
and a service-based measure (training, testing and referrals) which is also a measure of
output although some might describe it as a measure of multiple activities. The longterm rates of change for these measures are 1.0, 0.9, and 0 percent, respectively.
Not only do the service measurements vary by type of measure, but also the rate of
change of each measure may vary by the period examined. For example, the three
Employment Service measures show exactly the same rate of change (-3.3 percent) for
one peak-to-peak output cycle (1979-86), but for the preceding cycle (1974-79), the
rate of change was 4.6 percent for placements, 3.8 percent for referrals, and 2.1 percent
for the service-based measure (table 4).
Table 4. Comparison of rates of change for output measures for three services,
selected periods 1964-92
Measure
Rates of change
Mass transit
Unlinked passenger trips
Vehicle revenue miles
1967-92
.5
2.2
1970-81
1.5
2.6
1981-84
-0.4
.3
Unemployment Insurance
Program
Weeks compensated
1964-92
2.5
3.7
1967-72
2.2
4.9
1972-76
16.4
22.2
Employment Service
Placement
Referral
Service
1972-87
1.0
.9
.0
1974-79
4.6
3.8
2.1
1979-86
-3.3
-3.3
-3.3
SOURCE: Tables 31, 66, and 73
Introduction, Summary, and Conclusions—11
Do weighted and unweighted output trends differ? The desirability of differentiating
and weighting service output is discussed at several points in this bulletin. However, the
process often requires considerable data and effort to calculate weighted output indexes. The questions are, how much difference does it really make? And, if it makes a
difference, is it worth the effort?
Nine of the 10 service output measures presented in this bulletin are differentiated
and base-year weighted. Two, UI and ES, employ different outputs in the weighted and
unweighted measure, and for this reason are not considered further. The remaining
seven—electric power, natural gas, drinking water, mass transit, alcoholic beverage
sales, prisons, and juvenile institutions—use the same basic measure in the weighted
and unweighted output and can be directly compared.
Table 5. Comparison of rate of change of weighted and unweighted
output for seven services, selected periods 1967-92
Service and
period
Output
Unweighted
Weighted
Electric power ......................................
(1967-92)
3.5
3.6
Natural gas ..........................................
(1974-92)
-1.0
-0.7
Water supply ........................................
(1967-92)
1.7
1.8
Mass transit trips .................................
(1967-92)
.5
.5
Mass transit miles ................................
(1967-92)
2.2
2.2
Alcohol beverage sales .......................
(1967-92)
.3
-.1
State prisons ........................................
(1973-92)
7.7
7.8
Juvenile institutions .............................
(1971-92)
.2
.3
SOURCE: Tables 11, 18, 23, 31, 39, 49, and 61
Examination of the long-term weighted and unweighted trends for the seven services
show little difference (table 5). There is not much difference in the trends because there
was little change in the composition of the outputs in these services in the periods
studied. Only one service, State alcoholic beverage sales, shows a reversal in the sign
as a result of the weighting; this reflects the move by several States from retail to wholesale-only operations. Natural gas weighted and unweighted outputs also show a modest
difference, although not a reversal in the sign; this difference is due to the relative shift
in sales from industrial to residential service. The other five indexes show little or no
difference between the weighted and unweighted output indexes.
Introduction, Summary, and Conclusion— 12
Measurement outlook
This discussion has illustrated some of the possibilities and problems in computing
State and local government productivity indexes at the national level. The problems are
substantial and include both conceptual and data issues. However, the difficulties should
not be any worse for calculating State and local government productivity than for calculating private sector service industry labor productivity trends.
The two sectors produce many of the same services. There are literally dozens of
such services, ranging from electric power to alcoholic beverage sales to hospitals to
employment counseling. Not every government service has its private sector counterpart, but many do.
Furthermore, similar underlying economic, technical, and institutional forces are at
work in both government and private organizations. Environmental regulation, shifts in
energy prices, and imposition of water quality standards shape water utility productivity. Deregulation of energy prices, imposition of oil embargoes, and environmental
regulation affect the productivity of electric power and natural gas utilities. Environmental regulations, shifts in energy prices, and the demand for private automobile transportation dramatically reshaped the mass transit field. And demand for labor and new
technology is constantly remaking employment service operations.
Much of the past discussion on calculating government productivity has been entangled in questions of effectiveness and outcome. As long as the discussion is restricted to direct outputs, the solutions are generally as tractable for government services as they are for private sector services.
This is not to say that national labor productivity trends can be computed for every
State and local government service. Thorny problems exist in calculating State and
local government productivity, just as they do in the private sector. However, it should
be possible to compute State and local government labor productivity trends for many
services.12
Introduction, Summary, and Conclusions—13
Endnotes
1
Economic Report of the President, Washington: U.S. Government Printing Office, 1997, pp. 300-1, 338,
and 351.
2
As of 1995, more people were employed by Federal, State, and local government than in manufacturing.
Employment and Earnings, Vol 44, No. 2, February, 1997, p. 41.
3
Some State and local governments measure their productivity, or at least the productivity of parts of their
operations. However, national statistics, the focus of this report, are lacking.
4
National Research Council, Measurement and Interpretation of Productivity, Washington: National Academy of Sciences, 1979, pp. 9-10. See also, U.S. Congress, Joint Economic Committee, Productivity in the
Federal Government, Washington: Government Printing Office, 1979, p. 7; and U.S. General Accounting
Office, The Federal Role in Improving Productivity—Is the National Center for Productivity and Quality
of Working Life the Proper Mechanism? May, 1978, p. 45.
5
Walter L. Balk, Geert Bouckaert and Kevin M. Bronner, “Notes on the Theory and Practice of Government Productivity Improvement,” Public Productivity and Management Review, Vol. XIII, no., 2, Winter,
1989, p. 117.
6
Geert Bouckaert, “Public Productivity in Retrospective,” in Public Productivity Handbook, edited by
Marc Holzer, New York: Marcel Dekker, Inc., 1992, pp. 20-29. See also Darlene Forte, “Measuring Federal
Government Productivity,” in Handbook for Productivity Measurement and Improvement, edited by William F. Christopher and Carl G. Thor, Cambridge, Mass.: Productivity Press, 1993, pp. 7-3.1.
7
U.S. Bureau of Labor Statistics, Measuring Productivity in State and Local Government, Bulletin 2166,
Washington: U.S. Government Printing Office, December 1983; and U.S. Bureau of Labor Statistics, Productivity Measures for Selected Industries and Government Services, Bulletin 2440, Washington: U.S.
Government Printing Office, March, 1994.
8
Jesse Burkhead and Patrick J. Hennigan, “Productivity Analysis: A Search for Definition and Order,”
Public Administration Review, January/February, 1978, pp. 34-40, and Arie Halachmi and Marc Holzer,
”Introduction: Toward Strategic Perspectives on Public Productivity,” in Strategic Issues in Public Sector
Productivity, edited by Marc Holzer and Arie Halachmi, San Francisco: Jossey Bass, 1986, pp. 5-14.
9
Kent Kunze, Mary Jablonski, and Virginia Klarquist, “BLS Modernizes Industry Labor Productivity
Program,” Monthly Labor Review, July, 1995, pp. 3-12.
10
Jerome A. Mark, “Progress in Measuring Productivity in Government,” Monthly Labor Review, December, 1972, pp. 3-6, and Richard Murray, “Measuring Public-Sector Output: The Swedish Report,” pp. 51819, in Output Measurement in the Service Sectors, edited by Ziv Griliches, Chicago: The University of
Chicago Press, 1992.
11
Richard Murray, “Measuring Public-Sector Output: The Swedish Report,” pp. 531-33, in Output Measurement in the Service Sectors, edited by Ziv Griliches, Chicago: The University of Chicago Press, 1992.
12
Several research groups have conducted studies of government productivity. See Richard Murray, “Measuring Public-Sector Output: The Swedish Report,” pp. 517-42, in Output Measurement in the Service
Sectors, edited by Ziv Griliches, Chicago: The University of Chicago Press, 1992; R. Goudriaan, H. de
Groot, and F. van Tulder, “Public Sector Productivity: Recent Empirical Findings and Policy Applications,”
pp. 193-209, Proceedings of the 41st Congress of the International Institute of Public Finance, Detroit:
Wayne State University Press, 1987; and Malcolm Levitt and Michael Joyce, “Measuring Output and Productivity in Government,” Discussion Paper No. 108, London: National Institute of Economic and Social
Research, 1986.
Introduction, Summary, and Conclusion— 14
Chapter 2. Methodological
Considerations
U
nderlying the measurement of State and local government productivity is a
fundamental issue of methodology on which there is considerable disagreement. Not even the definitions and terminology are consistently applied. This
chapter discusses the basic conceptual issues and presents the approach used in this
bulletin.
The production framework and process
Government productivity measurement may be approached in one of two ways based
on how one measures output. One approach focuses on welfare aspects and considers
utility functions, indifference curves, and community satisfaction. The approach used
in this bulletin, however, focuses on production possibilities and considers production
frontiers and cost functions. This approach assumes that government production decisions can be modeled through a production function framework similar to that used in
private sector productivity analysis. It requires the general identification of outputs and
inputs but does not require detailed specification of a production function.1
The basic conceptual model is the following: Government draws on a series of inputs to undertake a series of activities which result in one or more outputs intended to
produce a series of desirable consequences. Inputs consist of labor, capital, and purchased materials and services. Activities are intermediate services or processes. Outputs are the final goods or services produced by the government. Consequences, which
are also known as outcomes and impacts, are the intended results of government action.2 A basic model of the production process is portrayed in chart 2.3
In its more sophisticated form, the model includes the citizen, who is a producer as
well as a consumer, and environmental and community conditions which affect service
production techniques (chart 3). In this model, consumers and the environmental setting are necessary parts of the production process, although their importance will depend on the service. They are likely to be much more important in education and
policing than in water supply, although even water supply will be affected by these
considerations.
For some government services, such as sanitation, the model can be applied in a
relatively straightforward manner. Sanitation organizations use laborers, drivers, trucks,
brooms, gas, and uniforms as inputs. The sanitation organizations use these inputs to
produce a series of activities such as sweeping streets, emptying litter cans, and picking
up residential trash. Output, in this case, might be the trash collected. The consequences should be cleaner streets and neighborhoods and fewer fire and health hazards.
The more sophisticated model also includes citizen inputs such as reporting of missed
collections to government, separating and preparing trash for recycling and disposal,
and carrying trash to the curb for pickup. It also includes environmental realities such
as the community's topography, household density, and climatic conditions.
For police services, inputs are patrol officers, police cars, communications equipment, and like items. Activities include recruiting and training police officers, and
taking calls from citizens. Outputs might include the amount of patrolling and the
number of arrests. The consequences of these actions should be a safer community.
The sophisticated model would include citizen behavior, such as reports to the police
and neighborhood watch activities.
Methodological Considerations—15
Chart 2. Basic model of the production process
Inputs
r1
r2
•
•
•
•
rn
Production
transformation
→
→
f
Outputs
Q1
Q2
•
•
•
•
Qm
Welfare
transformation
→
Consequences
→
F
C1
C2
•
•
•
•
Cs
Chart 3. Sophisticated model of the production process
Community
→
→
f
→
Welfare
Consequences
transformation
Outputs
Q1
Q2
·
·
·
·
Qm
Consumer
→
Government
→
Consumer
Production
transformation
→
F
→
Inputs
Community
→
C1
C2
·
·
·
·
Cs
For some services, there is general agreement as to what constitutes an activity, an
output, and a consequence, but for many services these are not always obvious (table 6).
For police and fire services, the intended consequences and activities are reasonably
clear-cut, but the outputs are not. For electric power, the activities and outputs are
reasonably clear-cut but the intended consequences are not. Public transit officials, for
example, often must provide service to a community and its residents. Their output is
the operating transit vehicle; the community's use of the service is a consequence. This
concept of transit service differs from the approach of the private sector transit manager, who is free, in most cases, to terminate unprofitable routes. In such cases, capacity provided is an activity while use of that capacity is the output.4 Although it is not
always easy to define outputs—or even to draw the line between a consequence and an
output, or between an output and an activity—they must be selected with care, service
by service and organization by organization. At the same time, there will often be
disagreement on what constitutes a government output and outcome.5
The service and the organization level will affect the activity, output, and consequence. The output of one organization may be an activity of another. Water meter
repair, for example, would be the output of the water utility repair shop but not of the
water utility. The output of a catalog unit of a public library would not be the output of
the library. This concern is similar to the intermediate output issue encountered in
private organizations. The outputs of an organization's personnel, data processing, budget, and communications units are inputs for other parts of the organization.
This study focuses on final organizational or government output, that is, service
provided to the community and its citizens. For the most part, it excludes consequences.
The focus is on the rate of change of final government output and the inputs (primarily
labor) which are used to produce the output.
Methodological Considerations—16
Table 6. Examples of steps in the production of selected government services
Service
or
function
Activity
Output
House offenders
Consequence
Corrections
Clothe inmates
Serve meals
Patrol cell blocks
Reduce crime
Protect society
Education
Conduct classes
Educate students
Give tests
Serve meals
Operate school buses
Increase literacy
Reduce unemployment
Fire
Maintain fire trucks
Train firefighters
Put out fires
Rescue citizens
Inspect property
for fire hazards
Reduce fire losses
Reduce fire deaths
Food stamps
Conduct interviews
Conduct audits
Issue stamps
Increase nutritional
levels
Library
Shelve books
Catalog books
Circulate books
Increase literacy
Street maintenance
Maintain trucks
Repair streets
Reduce traffic deaths
Reduce travel time
Water supply
Read meters
Deliver water
Improve community
Repair water mains health
Generate revenue to
support government
Measuring outputs
The specification and measurement of output is the single most troublesome problem in computing productivity. It is a more serious problem for government service
than for private sector organizations. Some of the problems are endemic to a specific
government activity; others are more general in nature. This section is concerned with
the latter. This bulletin addresses the issues of: Specifying the unit of measure, weighting the outputs, accounting for quality change, stipulating the criteria for selecting outputs, and examining the availability of output data.
Unit of measure. The basic output measure(s) of an organization should be a homogeneous physical unit. Furthermore, the measure should be related to the resources spent
in its production. Because of the problems in defining and measuring government output, a series of outputs needs to be examined and tested for each government service.
Street cleaning illustrates some of the problems and issues in selecting the appropriate unit of measure. Two commonly used measures of street cleaning output are cubic
yards of trash collected and curb miles swept. The curb miles swept will be about
proportional to the resources needed for sweeping. Also, the quality of service, such as
cleanliness, should be related to the output and input. On the other hand, cubic yards of
trash collected will not be as closely related to its resource inputs. In fact, the inverse is
likely: As streets are swept more frequently, the cubic yards of trash collected may
increase but at a much slower rate than labor inputs. The result is decreasing productivity. Curb miles is the preferred measure of output in this case.6
For most public sector output measures, physical quantities are used to calculate the
outputs. In the private sector, value data, such as revenue or sales, are usually used.
Methodological Considerations—17
In such cases, price changes are removed from the value data to obtain an index of real
output. Where the industry produces and sells a number of products, this approach
facilitates the calculation of output.7
The primary problem with using price-adjusted value as the measure of output in the
public sector, is that, in most cases, market prices are lacking. Without direct pricing,
estimating output in real terms is impossible.8 Exceptions are the enterprise services,
such as water and electric power, which are sold in the marketplace much as services
are sold in the private sector.9 They account for about 6 percent of total State and local
government employment. But even for the enterprise services, value is not always a
good measure of output because prices are administratively determined and many have
little relation to the costs of production. Transit, for example, is heavily subsidized, and
the subsidies are adjusted frequently. In such cases, physical measures are preferable.
A second issue in measuring government output is the degree of coverage within
each service. Most State and local governments produce multiple services. Some of
these are relatively easy to identify. Many sanitation departments, for example, sweep
streets, pick up trash, and remove abandoned cars, a set of easily identifiable services.
In other cases, the multiple services are not so easily identified. Some electric utilities,
for example, conduct energy audits as well as produce and deliver electricity.
There are two basic approaches to the construction of output indexes for multipleservice organizations. One is to identify each organizational product. For sanitation,
this might be household trash pickup (measured by tons removed), street sweeping
(measured by curb miles swept), and abandoned car removal (measured by the number
of cars removed). A separate index could be calculated for each product or service;
these, in turn, could be combined into a single sanitation index by using the appropriate
weights.
The other approach is to focus on the dominant output. The index for the dominant
or primary output would be used to represent the entire function. This approach is valid
when secondary outputs are unimportant (at least when the impact on productivity calculation is marginal) or when growth in uncovered output would about parallel the
growth in covered output. Kilowatt hours are usually used to measure the output of
electric power utilities. Other services such as energy audits and weatherization consume relatively few utility resources.
Whether single or multiple products are used to measure organizational output depends on whether the single product is representative of total output, the importance of
the single product and multiple products to decision makers, and data availability.
Another issue is accounting for only that part of the output actually produced during
the output cycle (e.g., year). This is not likely to be as significant an issue for State and
local government as it is in the private sector where the production of a single item, such
as an office building or a ship, may take several years to complete. Most State and local
government service outputs are started and completed in the same year. If the product
is not completed within the year, or in the accounting period used in the measurement,
an estimate must be made of what part of the final output is produced in each year so
that outputs and inputs match.
A fourth issue in specifying the unit of output is to include only the work produced
by the organization. Many government services are purchased from other governments
and private contractors. Many communities contract with private firms to remove household trash, care for juveniles, maintain street lights, and remove snow. Government
produced outputs must be separated from contracted outputs. In most cases, the separation is relatively straightforward assuming that data are available. Where a service is
provided partly by the government and partly by a private contractor, separation of the
output is likely to be more difficult; this mode of operation is becoming increasingly
popular.
Methodological Considerations—18
A parallel issue arises in contracting for intermediate services or activities such as
custodial or data processing services. In this case the outputs remain the same, and the
inputs reflect the shift from public to private or vice versa. Adjustments for these shifts
should be reflected in the inputs as they are in multifactor productivity calculations.
Output weights and aggregation. Calculation of a multiple-service output index or a
single service with multiple outputs requires aggregation of individual output
measurements. The traditional approach to private sector output calculations is to
apply revenue weights to the goods and services sold. For public sector services not
sold in the market place, the correct measure would be one that captures the marginal
social benefit of each output. Several different approaches have been suggested
including cost-effectiveness, cost-benefit, and political voting.10 Although conceptually attractive they are difficult to develop and are rarely used.
Fortunately, the production function has the cost function as its dual. Thus, it is
mathematically correct to use unit costs to weight outputs, and this approach is sometimes used in private sector calculations.11 It is particularly helpful with government
services where price weights are lacking. Another reason for using costs is that they are
more fully under the control of the government official, whereas outputs are often under
the control of others.12
Cost weights or unit costs are available for some government services, but in many
instances they are not. In such cases labor weights are often used as surrogates. For
most government services, unit labor inputs are a good proxy for unit cost because
labor comprises such a large part of government production costs. The weights used to
construct the State and local government output indexes presented in this bulletin are
unit labor or unit cost weights.
For historical and institutional reasons a 5-year chain-weighted system is used to
develop the indexes.13 The weights are revised periodically, usually every 5 years to
take advantage of the Bureau of Census, Census of Governments baseline survey. The
general equation used to weight and combine outputs is the following:
Output Index
=
∑L Q
∑L Q
0
i
0
0
where:
Lo = unit labor requirement in the base year
Qi and Qo = output quantities in the current and base years.
Quality of service. Quality change is a major issue in developing output indexes. For
many services it is an important, if not crucial, attribute of the output. It is of particular
concern in measuring government output. Increases in government expenditures are
often justified as quality improvements--streets are kept cleaner, snow is removed faster,
and police respond more rapidly to calls for assistance. Conversely, some feel that
productivity gains are made at the expense of quality. A study of New York City services published in 1976 concluded that output (quantity) had increased but performance
(quality) had deteriorated. In the case of the police department, 36 of 37 measures of
output quantity increased, but most measures of quality, such as the proportion of crimes
solved, decreased.14
There is considerable debate as to what changes in quality mean and how they should
be handled analytically.15 Of the two general approaches, one focuses on consumption,
the other on production. For consumption, a quality change is reflected in a change in
consumer utility; for production, a quality change is reflected in a change in resource
requirements. This bulletin is concerned with the latter.
When production and resource requirements change, adjustments need to be made
in the output (or input) index. A dated, but graphic, illustration of such a change is
found in the solid waste collection field. Several decades ago many cities shifted from
Methodological Considerations—19
backdoor to curbside collection. This required citizens to carry their trash to the street,
a task formerly performed by government collectors. In fact, what happened was that
the government introduced a new service or changed the level of service. The production process was modified and resource requirements were shifted as a result of this
change in service.
There are a number of less obvious, but equally important quality changes in government services in recent years. Some changes directly affected the consumer and others
affected the population as a whole. An example of the first type is drinking water where
major increases in treatment have improved its quality but raised its production cost.
Examples of the second type include sewage treatment, solid waste disposal, and electric power generation. In each of these cases, pollution control has raised the cost of
production, but resulted in environmental improvements.
Changes in the quantity or volume of production are sometimes viewed as quality
shifts. From the standpoint of the citizen, adding a branch library or a recreation center
may improve the quality of service because he or she will not have to travel so far to
reach a facility. From a production standpoint, it is simply an increase in output, that is,
an increase in the quantity or volume of production.
There are several ways to compensate for the change in quality. In the example of
the shift from backdoor to curbside trash collection, the input index could be adjusted
to include the work of the citizen (labor hours) in transporting the trash from the back
door to the curb. The output index would remain the same.
Another approach to a quality shift is to identify the new service and create a new
index. In this case, the new service is curbside collection; the old service, backdoor
collection. The two productivity indexes would be linked to create a single index.
The result of a change in quality on productivity measurement can vary depending
on the output measure chosen. For street cleaning, the more frequently the street is
cleaned, the cleaner the streets. If the output measure is curb miles swept, resource
requirements will remain about constant, productivity will remain about constant, and
the quality of service and output will be about proportional. This assumes that other
factors remain fairly constant. However, if the measure is cubic yards of trash collected, more frequent cleaning will result in a decrease in cubic yards collected with
each additional cleaning. In other words, resource requirements will increase and productivity will decrease, because of changes in the quality of service.
Clearly, quality should be examined function by function. For some functions, the
issues and variables, if not the solutions, are straightforward, while for others they are
complex and certainly not obvious.
Identifying the crucial quality considerations in State and local government services
is not easy. Although some research and discussion of quality and its measurement
have taken place over the past decade, little research has been done on the absolute or
relative effect of quality change on productivity, costs, and resource requirements. Lacking systematic research, the process has to be ad hoc.16
One approach for handling quality in State and local government productivity measurement is the following:
• Identify service output.
• List quality considerations for the output measure.
• Assess each quality factor for its potential impact on resource requirements.
• Create a quality index time series if the impact is potentially important.
• Track the quality index through time.
• Adjust the input index or link a new productivity index with the old index if the
quality index changes.
Methodological Considerations—20
Criteria for selecting outputs. To guide in the selection of State and local government
output measures, seven criteria are presented.17 The first four are essential; the last
three are desirable.
• Outputs must reflect the final product (service) of the organization. To determine pro
ductivity, the output must be the product or service leaving the organization, not an
intermediate product. Output must reflect the work rather than the consequences or
outcomes of the work.
• Outputs must be measurable. Absolute (cardinal) numbers are required. Arguments
that government services cannot be measured usually fail to distinguish among the
measurement of intermediate products, final outputs, and consequences of government service. Whether or not a service can be measured has to be considered function by function.
• Outputs must be repetitive. Construction of an output index requires a repetitive or
recurring set of services or products. The quality of service can change, because it
can be adjusted, but the basic service must be repetitive.
• Output data must be accurate and comparable. Much State and local governmentoutput
data currently collected, at least at the national level, are incomplete, inaccurate, and
inconsistent from period to period. Construction of a viable output index requires
reasonably accurate, comparable data. Comparability from one period to the next is
more important than absolute accuracy in preparing a time series.
• Output calculations should use existing data and data collection procedures. Two
issues are involved here--whether the records exist in State and local government,
and whether a procedure currently exists to collect national data. In either case,
existing data and data collection procedures should be used whenever possible, as
new procedures will likely be costly and time consuming.
• Outputs should be easily understood. An index that is simple and easily understood
is most likely to be accepted, supported, and used. Esoteric measures and complex
quality adjustments of dubious reliability should be avoided.
• Output units should reflect the resources spent in their production.
Availability of data. Output data, and their availability, play a major role in selecting the
services to be measured. It also plays a role in the specific outputs measured. It is a
limiting factor for national calculations where output data for even the most straight
forward service, such as solid waste management, are lacking. Data are more readily
available at the State and local level. Many governments routinely prepare statistical
tabulations and performance reports from which output indexes can be constructed.
However, even for the individual government, output data are often lacking.
Measuring inputs
This section discusses the number and type of factor inputs used to measure productivity. It also presents the labor measures most often used and reviews some of the
questions surrounding these measures. It presents the criteria to be used to select inputs,
and discusses data currently collected that might be used to calculate national labor
indexes for State and local government.
Number and type of inputs. The number and type of inputs is often used to characterize
productivity measures. A common characterization is single factor and multifactor.18
In reality, there is a continuum of inputs or factors. The number and type of resource
inputs used should reflect the use to which the measure is put.
A single factor productivity measure, the most common type, relates one resource,
most often labor, to output. It does not measure, however, the specific contribution of
the factor to output. Rather, it expresses the joint effect of interrelated influences, such
as management, technology, and regulation, as well as changes in other inputs relative
Methodological Considerations—21
to the measured input, on overall output.
Multifactor productivity relates two or more inputs to output and also reflects the
joint effect of many influences. However, it eliminates the effect of the substitution of
one factor for another on overall production. Examples of multifactor inputs are labor
and capital; labor, capital, and energy; or labor, capital, energy, and materials.
This bulletin focuses on single factor productivity, specifically labor productivity,
for several reasons. First, labor is of primary importance in public policy issues. Labor
compensation constitutes about 40 percent of all State and local government expenditures (capital and operating), but about 60 percent of all operating expenditures. Its
importance varies by type of service (table 7). For example, compensation constituted
90 percent of the expenditures for police programs but 9 percent for State alcoholic
beverage sales in 1992. Second, labor is relatively easy to calculate when compared to
other factors of production. Third, labor data have been collected for many years and
generally are the most accessible of State and local government factor inputs. Fourth,
labor indexes are calculated for many parts of the private sector and for some foreign
countries. State and local government labor-based indexes permit comparisons with
these other sectors and institutions.
Two labor measures that are commonly used to measure private sector productivity
are the number of hours and the number of persons. These two measures impart very
different information and for this reason both are calculated. The preferred measure is
the number of hours because this measure reflects the time worked to produce the outputs.
Hours are the preferred measure for government too, but governmental units do not
usually collect these data. Instead, most governments collect and use the number of
full-time-equivalent employees.
Full-time-equivalent employees. A full-time-equivalent employee, or an employee year,
usually equals 2,080 hours (40 hours per week times 52 weeks per year) and includes
all paid time such as overtime, vacation, holidays, and sick leave. Part-time employment is usually converted to a full-time-equivalent basis, such as the number of parttime employees whose hours add up to 2,080 equals one full-time-equivalent employee.
Seasonal employment is computed in the same way. Overtime should be handled in the
same manner but frequently is simply ignored.
Many practical problems arise in measuring the paid time of State and local government workers. Should standby time of police, fire, and public works officials who are
home, but subject to call, be included? Sometimes employees are paid for standby time
but more often they are not. What about employees who are paid by the task, such as
collecting trash on a specified route? When they finish the task they are permitted to go
home. Regardless of the actual time worked, the employees are paid for a fixed, previously agreed upon time.The converse--time worked but not paid--also needs to be considered. Many employees, including managers, teachers, and coaches, work hours for
which they are not paid. Conceptually, this time should be counted too, for if it is
increasing (or decreasing), productivity trends will be overstated (or understated) if not
included. Most of these issues relate to a specific government function or service and
should be addressed in that context.
Number of employees. This index simply counts the number of employees who produced the output without regard for the time each employee worked. A part-time employee is counted the same as a full-time employee. An index of the number of employees understates the change in labor input when the time worked per person increases,
such as overtime, and overstates the change in labor input when the time worked per
person decreases. The greatest divergence between an FTE index and an employee
index probably occurs when part-time employment increases or decreases.
Most State and local governments use part-time employees extensively. In October
1992, they employed 12.0 million full-time and about 3.7 million part-time workers (24
percent of the total). However, part-time employment varies substantially by service.
Methodological Considerations—22
Table 7. State and local government labor compensation as a percent of total operating expenditures for selected functions, fiscal
year 1992
Percent
Function
All functions ...............................................
58.5
Police ............................................................
Fire ................................................................
Education ......................................................
Corrections ....................................................
Libraries ........................................................
90.4
89.5
82.8
76.5
66.0
Hospitals .......................................................
Finance administration ..................................
Highways .......................................................
Transit ...........................................................
Natural resources ..........................................
65.9
62.2
61.3
60.6
58.0
Parks and recreation .....................................
Water transport .............................................
Health ............................................................
Air transportation ..........................................
Water supply .................................................
54.3
45.9
42.1
41.0
39.1
Sewerage ......................................................
Solid waste management ..............................
Housing and community development ..........
Electric power ...............................................
Gas ...............................................................
38.6
34.5
26.9
17.6
14.3
Welfare ..........................................................
Liquor sales ..................................................
11.6
8.8
SOURCE: Computed from data taken from Bureau of Census Government
Finances: 1991-92; Public Employment: 1992; and unpublished data from Bureau of Economic Analysis.
In 1992, 4 percent of natural gas utility employees worked part time, in corrections
it was also 4 percent, and it was 6 percent in sewerage. But the work forces of local
libraries and higher education consisted of 47 percent and 52 percent, respectively, of
part-time workers.19
Seasonal employment such as for snow removal, leaf pickup, park maintenance, and
swimming pool operation can create measurement problems when calculating an index
of the number of employees. The primary problem is the period of coverage. Employee counts are commonly published for one date such as December 31 each year, but
an employee count on July 30 or October 30 may be quite different because of seasonal
variations. To overcome the problem of seasonal employment, the preferred approach
is to use a weekly or monthly average of the number of employees to calculate the
index.
Comparison of the two approaches. The FTE employee and total employment labor
indexes could produce very different trends and movements depending on the period
examined. For this reason, it is preferable to calculate both indexes but this is not
always possible because of the lack of data.
Whether the two approaches to State and local government labor measurement would
produce markedly different labor trends in the public sector is not known. Private
sector labor trends for hours and employees differ, but only slightly over the long run.
Several government services, as noted above, employ a large number of part-time workers, and could produce very different employee counts as a result. Whether these would
Methodological Considerations—23
affect labor trends is another question. As long as the ratio between the number of fulltime and part-time employees remains constant, the trends of the two series are not
affected.
The two indexes did not vary greatly for the services examined, that is, when both
could be calculated. But only a few services were examined and for most of these there
was data to compute only one type of index. Furthermore, services employing large
numbers of part-time workers such as higher education, recreation, or fire fighting were
not examined.
Volunteers. Most governments use volunteers, although the extent of use varies considerably.20 They are common in services such as fire, education, hospitals, museums, and
recreation but are relatively rare in others such as public utilities.
Conceptually, a labor index should include volunteer participation. These workers
contribute to output just as do paid employees. However, in the real world of data
collection and measurement, identifying volunteer labor input or output is extremely
difficult. Probably in no more than four or five State and local government services are
volunteers even potentially important. If the ratio between paid and volunteer labor
remains constant in these areas, their inclusion or exclusion will not affect labor trends.
Further, records on volunteers are almost nonexistent in State and local government.
Productivity calculations usually include the output generated by volunteers but not
their inputs.
Construction employees. Force account,21 or construction employees, are used in the
production of a limited number of government services, primarily in the public works
area. Because of the focus on recurring service products and the resources used to
produce these services, only those force account employees are included who perform
routine maintenance and repair. This requires separating force account employees into
two groups: those who are involved in new construction and those who perform routine
maintenance and repair. In the case of water supply, only those workers are included
who clean existing mains and repair water main breaks. Those who extend water mains
to new subdivisions are not included.
For the most part, national government employment data sets do not separate force
account workers by the task performed. Thus, it is impossible to identify force account
workers, much less to separate them between new construction and maintenance. An
exception is the mass transit data collected by the U.S. Department of Transportation as
discussed in chapter 3. Fortunately, force account employees are not an issue for most
government services, and for those using force account employees, new construction is
normally contracted. However, there will be some services, in particular utilities, where
force account labor is included in employee counts and it is impossible to identify it. If
the ratio remains the same between the force account and other employees their inclusion, however, will not effect the labor trends index.
Changes in work force composition. Labor is often treated as a homogeneous input in
productivity calculations although clearly it is not. Depending on the mix, labor inputs
can produce very different levels of output. If the mix changes, the level of output can
be affected. For example, police departments increasingly require new recruits to have
some college education. The rationale behind the requirement is the creation of a police force that can better deal with the public and with today's complex society.
In theory, an increase in police education will increase police output or the quality of
output, that is, an improvement in the quantity and/or quality of police output. However, an increase in output resulting from additional education is not an increase in
productivity but an improvement in labor input, i.e., a shift in the composition of the
labor input.
The method generally used to adjust for changes in labor force composition is pay
differentiated by skill mix (e.g., education and experience). This requires information
on the change in pay by skill through time, data that are not readily available from State
Methodological Considerations—24
and local governments. Although often discussed in private sector productivity measurement, changes in labor force composition for State and local government are of
secondary concern at this time and are not considered further in this bulletin.22
Criteria for selecting inputs. Five criteria are suggested for identifying input data,
specifically labor inputs. Those essential for productivity measurement are presented
first, those less so follow.
• Inputs must match output. Calculation of productivity requires that the resources
applied match the measured organizational output. For organizations with multiple
multiple outputs, like the typical city government, this requires careful identification
of resources used to produce the outputs.
• Inputs must be measurable. Absolute numbers are required.
• Inputs must be accurate and comparable. Much of the labor data collected on State
and local government operations is inaccurate and inconsistent from period to perriod. Comparability is more important than absolute accuracy. Data analysis should
be part of the construction of any index.
• Input calculations should use existing data. New data collection procedures will
likely be time consuming and costly to develop and maintain, and burdensome for
those providing the data. Existing data and collection procedures should be used
whenever possible.
• Inputs should be easily understood. General acceptance, support, and use of an
index are more likely if the construction is straightforward and easily understood.
This is one reason that labor indexes are widely used.
Availability and accuracy of labor statistics. Most State and local governments collect
labor statistics for use in their day-to-day operations. Two types of labor measures, the
number of full-time-equivalent employees and the number of employees, are often collected. Most State and local governments should be able to prepare labor indexes by
function and for the government as a whole.
Preparation of national or regional labor indexes is not as straightforward. Some
labor data are collected and published by function by trade associations, public interest
groups, and Federal agencies. The International City County Management Association, the American Public Works Association, the American Water Works Association,
and others routinely collect statistics on public employment for specific functions and
sometimes for government as a whole. Federal agencies such as the Department of
Labor, the Department of Justice, and the Department of Transportation sometimes
collect data on the number of State and local government employees for the programs
they fund and coordinate.
In addition, there are four sources of national State and local government employment statistics: The Census of Governments; the Current Population Survey (CPS); the
Current Employment Statistics survey (CES-790); and the Unemployment Insurance
reports (ES 202). The characteristics of these surveys are briefly discussed in the following paragraphs.
Census of Governments, the best known, and probably the most widely used national statistics on State and local government employment, is produced by the U.S.
Bureau of the Census. Sample data are collected and published annually. Every 5 years
(years ending in 2 and 7), the Census Bureau takes a complete enumeration and publishes the results. Statistics are collected and published on the number of employees
(full time and part time) and the number of full-time-equivalent employees. Also, the
Census of Governments' employment series includes data on salaries and wages.
Methodological Considerations—25
There are several problems and potential problems in using these employment statistics to calculate government productivity. First, the statistics are for a single month,
October, of each year. Second, the information is not available until months, and sometimes years after the reference date. Third, the functional classification system used by
Census is very broad, e.g., police, fire, and employment security. Fourth, each government defines who is to be covered under a function in its own way and this definition
often differs from government to government. Furthermore, the definition can change
over time.
The Current Population Survey (CPS) collects data primarily to calculate monthly
employment and unemployment statistics. A number of other statistics, such as hours
worked and pay, are collected too. The strengths of this survey for productivity measurement are its timeliness and the information on hours worked. However, the CPS
presents a number of problems for State and local government productivity measurement. First, it contains no information on services or functions. Second, it is impossible to separate the employment data by type of government. And third, the employment counts for State and local government as taken from the CPS are markedly different from those obtained from other sources. In short, the CPS is not suitable for State
and local government productivity studies.
The Current Employment Statistics survey (CES 790) collects data monthly from
establishments in nonagricultural industries and government on the number of employees, average hours worked, and average hourly and weekly earnings. About half of all
State and local government employment is covered. Employment statistics are broken
down into eight functional areas for State government and into seven areas for local
government. The advantage of the CES 790 data for productivity measurement is its
timeliness; preliminary data are published about 3 weeks after reporting. However, the
CES survey has two deficiencies. First, statistics are not available for many government functions. Second, and most important, coding of the data by government function is poor, albeit improving.
Unemployment insurance has covered all State and local government employees
since 1978. As a result, State and local governments record monthly employment and
wages and report the data quarterly to the U.S. Unemployment Insurance program (ES202). Because the ES 202 is linked to financial reports, it provides the most accurate
statistics available on the number of persons employed by State and local government,
by State, county, and metropolitan statistical area.
Although the ES 202 reports are comprehensive, they lack detail and provide inadequate division by function--most employees are assigned to the general government
category. The attraction of the ES 202 report lies not in its current form but in its
potential if the coding by function were improved.
Other issues
Conclusion. Four observations are drawn from this examination concerning the measurement of government labor inputs for productivity calculations. First, no single data
source is likely to be entirely acceptable. Second, major errors are likely in each data
series. Third, viable labor-based input indexes require detailed data comparison and
adjustment irrespective of which data set is used. Finally, the best labor data sources
for productivity calculations are the special surveys that collect output and input data in
the same instrument.
This section discusses issues of productivity comparisons, frequency of measurement, geographic coverage, period coverage, and service definitions, which were only
briefly noted in the preceding sections.
Productivity comparisons—levels and trends. Underlying all productivity measurement is comparison: through time, of producing units, and of producing units through
time. Most private sector productivity measurements are time comparisons, such as,
"Productivity increased by x percent between 1973 and 1995." Trends are routinely
computed for industries, for individual countries, and for groups of countries.
Methodological Considerations—26
Similarly, State and local government productivity trends might be calculated and
stated as, "Municipal electric power productivity increased by x percent between 198792." Productivity trends could also be computed, for example, for mass transit in New
York City between 1967 and 1992, or for the Unemployment Insurance program in the
Southern States between 1964 and 1992.
Some productivity measures focus on absolute levels: "Each employee produces, on
average, x tons of steel or y cars per year." Local government measures might be, "x
tons of trash collected per employee" or "y miles of street swept per employee." Comparisons of levels could be made between jurisdictions or regions, or with the Nation as
a whole.
Trends and levels complement each other, thus a true picture of productivity requires examination of both. A city service might have a low level of productivity relative to other operations, but a high rate of productivity change, or vice versa. However,
the data and analyses required to compute levels are much more demanding than those
required to calculate trends. This bulletin focuses on productivity trends.
Frequency of measurement. National productivity trends are normally calculated annually, although some estimates are produced quarterly. The periodicity of calculation
is largely a function of data availability.
Benefits that would accrue from more frequent measurement are not obvious. Furthermore, monthly or quarterly productivity movements might not even be detectable
for most State and local government services. In addition, seasonal adjustments would
have to be made if quarterly or monthly calculations are to be useful. This requires
further knowledge about seasonal fluctuations.
Finally, some State and local government services, such as education, have outputs
that require more than one month or one quarter to produce and, thus, to measure. For
all the above reasons, the data presented here are annual statistics.
Geographic coverage. For purposes of this bulletin, State and local government includes the 50 States, the District of Columbia, and all cities, counties, special districts,
townships, and school districts. These jurisdictions comprised about 85,000 State and
local governmental units in 1992.
Geographic data source coverage is not always consistent. Some sources include
trust territories as well as the 50 States; others include only the larger jurisdictions.
Whether trust territories are included is probably not important because they are such a
small part of the total. However, by focusing on the larger jurisdictions, bias may be
introduced into productivity calculations.
The size as well as the location of a jurisdiction can affect its productivity. Thus, it
is important for national statistics to have complete coverage or, if sample data are
used, to have a representative sample. Some government services, such as water, sewerage, electric power, and refuse collection, benefit from economies of scale.23 Other
services, such as police and recreation, evidently do not. Topography and climatic
conditions affect garbage collection and street repair. The focus here is on national
statistics.
Time period coverage. Two time dimensions need to be considered in calculating productivity, the number of periods to be covered by the index (e.g., weeks, months, or
years) and the length of the time period (e.g., 12 months or 52 weeks for each year).
The time period covered by a productivity index, and the beginning and ending period,
can have a marked effect on the overall rate of change. Generally, the longer the time
span, the less important the beginning and ending years. Also, a longer period is usually more representative of long-term trends.
Cyclical fluctuations can affect productivity trends. Such fluctuations occur most
often when inputs lag the change in outputs. Unemployment Insurance outputs, for
example, parallel the unemployment cycle, and labor inputs usually lag behind changes
in outputs, as discussed later in this bulletin. The results are productivity indexes that
Methodological Considerations—27
shift significantly depending on the years included in the index. To avoid arbitrary
cutoff dates, and to reflect long-term trends more accurately, average annual growth
rates are usually calculated from output peak to output peak.
Structural shifts in the economy or changes in legislation can also influence calculations of long-term rates of productivity change. State and local government electric
power productivity dipped markedly with the increase in energy prices in 1973 just as it
did in the private sector. Revisions in State Alcoholic Beverage Control laws have had
dramatic effects on State alcoholic beverage store sales and productivity.
The second time-related issue is whether the calendar, or special year, such as fiscal
or program, should be used in productivity calculations. Most private sector productivity indexes are based on the calendar year. The recently terminated Federal Government productivity measurement system, however, used the Federal Government's fiscal
year of October 1 to September 30.
The question is more complicated in the case of State and local government productivity calculations. For a single government, or a group of governments with the same
fiscal year, there should be no problem. Data will cover the same period through time.
The Bureau of the Census, for example, asks all State and local governments to use the
July 1-June 30 fiscal year in reporting financial data. However, State and local government fiscal years vary. The U.S. Department of Transportation collects data from over
300 transit systems with fiscal years ending on March 31, April 30, June 30, September
30, and December 31. The U.S. Department of Energy electric power data are reported
by calendar year and the U.S. Department of Labor collects unemployment insurance
statistics from the States by the Federal fiscal year.
The closing month of the "productivity" year is not important for trend determinations, but the same month should be used each year. More importantly, the inputs and
outputs should cover the same period.
Service specifications and definitions. Definitions of services vary among governments
and through time. Public works, for example, may be specified as a single unit; may be
broken into major components such as sanitation, water supply, and street maintenance;
or may be divided into subservices such as solid waste residential collection, street
sweeping, street flushing, and so forth.
For single measurements or studies, definitions can usually be adjusted to meet analytic needs and data availability. For preparation of a national, aggregate productivity
index, a formal classification system is needed.
Most of the summary data in this bulletin are based on the classification system and
definitions of the Census of Governments (see appendix A). The Bureau of the Census,
State and local governments, and the research community have used this structure for
years.
However, even this classification system has several deficiencies for productivity
measurement. First, the service categories are very broad. Second, governments differ
in the manner in which they structure, and thus report, their operations. For example,
the functions assigned to police departments or State alcoholic beverage control agencies vary from jurisdiction to jurisdiction.
The Standard Industrial Classification (SIC) system, in contrast to the Census of
Government system, is very detailed. It includes all goods and services produced by
private and public establishments. State and local government is a small part of the SIC
(appendix A lists State and local government services included). Unfortunately, SIC
categories are not widely used in State and local government data collection and analysis.
Both the SIC and Census of Governments classification systems should be helpful in
structuring State and local government productivity analysis, collecting data, and making comparisons. However, for some services, such as the Unemployment Insurance
and the Employment Service, neither classification scheme is sufficiently detailed, and
further specification is necessary.
Methodological Considerations—28
The productivity index
Productivity indexes assume a variety of forms. The one used here relates an output
index to a labor-input index. The type of output index used is a base-period laborweighted composite index, with weights that change every 5 years. BLS used this form
for years in its industry labor productivity measurement program.24 It was used because of past computations, availability of data with which to make the computations,
simplicity of computation, and its potential use in decision-making. The resulting productivity index was developed to estimate the amount of labor required to produce a
given volume of goods and services. This index is especially germane for calculating
government productivity where an organization must respond to a request for its services, and the interest of the decision maker is the amount of labor needed to respond to
the request.
The mathematics of the base-period labor-weighted composite index is simple and
straightforward.25 The output index compares the quantity of services in the current
year with the quantity in the reference year (such as 1987). For each year, the quantity
of service is an aggregate computed by weighting detailed services with the number of
full-time-equivalent employee years expended per unit produced.
The index of productivity may be expressed as follows:
Pi =
Qi Hi
÷
Q0 H 0
where:
Pi = the index of productivity or output per FTE employee year in year i
Qi = the output quantity in year i
Q0 = the output quantity in year 0 (the reference
year)
Hi = aggregate FTE employment in year i
H0 = aggregate FTE employment in year 0
Government organizations that produce a single, uniform output, measure output by
counting the number of units produced. Among government organizations producing a
number of different services—the more typical case—output is computed with the baseperiod labor-weighted composite output index. The following form is used:
q 
Qi
= ∑ w0  i 
Q0
 q0 
where: w0 =
ho
H0
qi = the quantity of an individual service in year i
q0 = the quantity of an individual service in year 0, and
h0= the amount of labor expended in year 0 to produce an individual service.
Methodological Considerations—29
The output index can also be written in this form:
Qi
=
Q0
∑l q
∑l q
0
i
0
0
where l0 is the unit labor requirement of an individual service in the base period
(which is calculated as h0/q0).
Notice that the labor input measures, H0 and Hi, can be expressed as follows:
H 0 = ∑ l0 q 0 and
H i = ∑ li q i
where li is the unit labor requirement of an individual service in year i. Therefore the
productivity index can be written as
Pi =
Qi Hi
÷
=
Q0 H 0
∑l q
∑l q
0
i
0
0
÷
∑l q
∑l q
i
i
0
0
=
∑l q
∑l q
0
i
i
i
This expression shows that the productivity index compares the number of FTE
employee years that would have been required in the base period in order to yield the
services actually produced in year i to the number of FTE employee years actually used
in year i.
Methodological Considerations—30
Endnotes
1
Reino T. Hjerppe, “The Measurement of Real Output of Public Sector Services,” The Review of
Income and Wealth, June 1980, p. 239.
2
For further exposition of this model, see: D.F. Bradford, R.A. Malt, and W.E. Oates, “The
Rising Cost of Local Public Services: Some Evidence and Reflections,” National Tax Journal,
Vol. XXII, No. 2 (June 1969), pp. 185-202. See also Gordon P. Whitaker and others, Basic
Issues in Police Performance, Washington: U.S. National Institute of Justice, 1982, pp. 92-123.
3
Reino T. Hjerppe, “The Measurement of Real Output of Public Sector Services,” The Review of
Income and Wealth, June 1980, p. 240.
4
Gordon J. Fielding and others, Development of Performance Indicators for Transit, Irvine,
California: University of California, 1977, and Anthony R. Tomazinis, Productivity, Efficiency,
and Quality in Urban Transportation Systems. Lexington, Massachusetts: D.C. Heath and Company, 1975.
5
James Q. Wilson categorizes government organizations according to the difficulty in specifying
their outputs and outcomes. For some agencies, both can be specified, for others neither can be
enumerated, and for the remaining units one or the other, but not both can be identified. See
James Q. Wilson, Bureaucracy: What Government Agencies Do and Why They Do It, New York:
Basic Books, Inc., 1989, pp. 158-71.
6
U.S. Bureau of Labor Statistics, Measuring Productivity in State and Local Government (U.S.
Department of Labor, Bureau of Labor Statistics), Bulletin 2166, December 1983, pp. 52-59.
7
U.S. Bureau of Labor Statistics, BLS Handbook of Methods, Bulletin 2414, Washington: U.S.
Government Printing Office, September, 1992, p. 92. (Note: This edition of the Handbook has
been superseded by Bulletin 2490 but Bulletin 2414 is cited here and elsewhere in this publication because it describes the methods and procedures used to prepare this bulletin.)
8
John P. Ross and Jesse Burkhead, Productivity in the Local Government Sector, Lexington,
Massachusetts, Lexington Books, 1974, p. 35.
9
Government enterprise service outputs are treated as private sector products in the U.S. national income and product accounts.
10
Richard Murray, “Measuring Public-Sector Output: The Swedish Report,” in Output Measurement in the Service Sectors, edited by Zvi Griliches, Chicago: The University of Chicago Press,
1992, pp. 523-25.
11
Robert W. Crandall and Jonathan Galst, “Productivity Growth in the Telephone Industry Since
1984,” in The Service Productivity and Quality Challenge, edited by Patrick T. Harker, Boston:
Kluwer Academic Publishers, 1995, pp. 391-405.
12
This issue is discussed further in chapter 4. Also, see Peter Schmidt and Ann D. Witte, An
Economic Analysis of Crime and Justice: Theory, Methods, and Applications, New York: Academic Press, Inc., pp. 263-80.
13
Comparison of private sector output calculations using a 5-year chain-weighted system and a
revenue weighted procedure (Tornqvist indexes) shows little difference in the long-term growth
rates for most industries. Kent Kunze, Mary Jablonski and Virginia Klarquist, “BLS Modernizes Industry Labor Productivity Program,” Monthly Labor Review, July, 1995, pp. 3-12.
14
David Greytak, Donald Phares, and Elaine Morely, Municipal Output and Performance in
New York City, Lexington, Massachusetts: Lexington Books, 1976, p. 110.
Methodological Considerations—31
15
Franklin M. Fisher and Karl Shell, The Economic Theory of Price Indices, New York: Academic Press, 1972. Jack E. Triplett, “Robert Gordon’s Approach to Price Measurement,” BLS
Working Paper 101, Washington: U.S. Bureau of Labor Statistics, April, 1980; and Mark
Sherwood, “Difficulties in the Measurement of Service Outputs,” Monthly Labor Review, March,
1994, pp. 11-19.
16
A number of private sector productivity studies have used the hedonic methods to identify and
correct for changes in the quality of products and services. So far as is known this approach has
not been used in the measurement of government services.
17
For a slightly different list, see Brian Usilaner and Edwin Soniat, “Productivity Measurement,”
in Productivity Improvement Handbook, edited by George Washnis, New York: John Wiley,
1981, p. 95. For a more extensive discussion of criteria to be used in developing performance
measurement systems see, Geert Bouckaert, “Measurement and Meaningful Management,” Public Productivity and Management Review, Vol. XVII, no. 1, Fall 1993, pp. 31-43.
18
Multifactor productivity is often referred to as total factor productivity. But because a variety
of factors are referred to it is more appropriate to use multifactor.
19
U.S. Bureau of the Census, Public Employment: 1992, Series GE/92-1, Washington: Government Printing Office, 1994, p. 5.
20
Martha A, Shulman, “Alternative Approaches for Delivering Public Services,” Urban Data
Service Reports, Vol. 14, No. 10, Washington: International City Management Association, October, 1982, pp. 8-9.
21
These are construction workers employed by an organization not primarily engaged in the
construction trades.
22
U.S. Bureau of Labor Statistics, Labor Composition and U.S. Productivity Growth, 1948-90,
Bulletin 2426, U.S. Department of Labor, December 1993.
23
William B. Fox, Size Economies in Local Government Services: A Review, Washington: U.S.
Department of Agriculture, Economics Statistics and Cooperative Service, 1980.
24
BLS recently switched its industry computations to the Tornqvist index, an index that is more
in keeping with current economic theory. See Kent Kunze, Mary Jablonski, and Virginia Klarquist,
“BLS Modernizes Industry Labor Productivity Program,” Monthly Labor Review, July 1995, pp.
3-12.
25
U.S. Bureau of Labor Statistics, BLS Handbook of Methods, Bulletin 2414, U.S. Department
of Labor, September, 1992, pp. 89-90 For further discussion of the different types of measures
see W.E. Diewert, “The Measurement of Productivity,” Bulletin of Economic Research, 1992,
pp. 163-98. See also S. Grosskopf, “Efficiency and Productivity,” chapter 4 in Harold O. Fried,
C.A. Knox Lovell, and Shelton S. Schmidt, The Measurement of Productive Efficiency, New
York: Oxford University Press, 1993, pp. 160-93.
Methodological Considerations—32
Chapter 3. Enterprise Services
This chapter discusses the measurement of State and local government enterprise
services, presenting indexes for five of them: Electric power, natural gas, water supply,
mass transit, and alcoholic beverage sales. Enterprise services are distinguished from
other State and local government services by their similarity to private sector operations. In most cases their outputs are sold in the private sector, they have a great deal of
autonomy in their operations, and their fees and charges cover their operating expenses.
There are a variety of State and local government enterprise services, the precise
number depending on how the services are categorized and counted. The National
Income and Product Accounts (NIPA) prepared by the U.S. Department of Commerce
use the following breakout:1
Airport terminals
Electricity supply
Housing and urban renewal
Liquor stores
Natural gas
Miscellaneous commercial activities
Sewerage
Toll highways
Transit
Water supply
Water terminals
Miscellaneous commercial activities include lotteries, off-track betting, parking, and
other miscellany.
The level of fees and charges and the percent of operating costs covered by enterprise service sales vary by service. Some, such as electric power and natural gas, are
covered entirely by sales. Others, such as mass transit, are heavily subsidized by general taxes.
Each enterprise service is briefly discussed in this chapter. The chapter also looks at
potential output measures and data to calculate the output indexes, examines labor input data, and calculates several productivity indexes. The specific approach and time
period covered vary by service, depending on data availability. The concluding section
summarizes some of the lessons learned in measuring enterprise service productivity.
Electric Power
Electric utilities are a good starting place for a discussion of the measurement of
State and local government enterprise service labor productivity. They are easily identified, they have a readily measurable set of outputs, and they report annually to the
Federal Government. Furthermore, productivity indexes have long been calculated for
private, cooperative, and government electric utilities. Thus, there is a large analytical
and institutional base of knowledge on which to build a discussion.2
Institutional setting
Electric utilities can be separated into three basic types, based on the type of ownership—private, cooperative, and government. The private or investor-owned utilities
account for about three-fourths of all production and sales. The 262 privately-owned
utilities sold 76 percent of the Nation’s kilowatt hours, served 76 percent of the customers, and owned 75 percent of the Nation’s electric plant and equipment capacity in 1992
(table 8).
Enterprise Service—33
Table 8. Percent distribution of kilowatt hours sold, customers served, and
installed capacity owned by type of utility ownership, 1992
Type of ownership
Total ................................................
Private (investor) .............................
Cooperative (REA) ..........................
Government ....................................
Federal .........................................
State and local .............................
Kilowatt hour Customers
served
sales
100.0
76.4
7.5
16.1
1.8
14.3
100.0
75.7
10.6
13.7
(1)
13.7
Installed
capacity
(kwh)
100.0
75.1
4.4
20.5
9.0
11.5
1
Less than 0.05 percent.
SOURCE: Public Power, January/February, 1994, pp. 72-74, and U.S. Energy Informamation Administration, Financial Statistics of Major U.S. Publicly Owned Electric Utilities,
1992, Washington: U.S. Department of Energy, January 1994, p. 3.
The second type consists of the user-owned Rural Electrification Administration
(REA) cooperatives, which expanded dramatically into the rural areas in the 1930s
under the sponsorship of the Federal Government. The 943 cooperative systems sold
about 8 percent of the Nation’s electricity, served about 11 percent of its customers, and
owned about 4 percent of utility plant and equipment in 1992.
The third type, the government-owned utility, includes two basic types of utilities,
Federal and State and local. The Federal Government is primarily a generator and
wholesaler of electric power. It sells about 2 percent of the Nation’s kilowatt hours to
final customers which accounts for less than 1 percent of the final users, but owns about
9 percent of the Nation’s installed generating capacity.
State and local electric power systems, sometimes known as “municipal systems,”
include State, special district, city, and county operations.3 The 2,017 municipal systems covered by the BLS productivity indexes presented in this bulletin accounted for
about 14 percent of the Nation’s kilowatt sales, 14 percent of its customers, and 12
percent of electric plant installed capacity in 1992.
Kilowatt hour sales of the major State and local systems in 1992 went to residential
users (34 percent), commercial users (25 percent), industrial users (35 percent) and
“other” users such as other public power authorities, railroads, and highway and street
lighting authorities (6 percent).4 Generating capacity was divided among coal (32 percent), natural gas (22 percent), hydroelectric (22 percent), nuclear (13 percent), oil (10
percent), and other (1 percent).5
The gross revenue of the State and local government systems was $31.0 billion in
1992. Expenditures, including capital investment, were $32.0 billion. Operations accounted for $22.7 billion. Compensation (wages, salaries, and benefits) accounted for
about $3.8 billion or 17 percent of total operations. Fuel, materials, supplies, and purchased power accounted for the rest (table 9).
State and local systems are scattered throughout the United States. The District of
Columbia and Hawaii are the only jurisdictions which have no State or local government power systems.6 California led the list with 13,418 full-time-equivalent employees and $3,211 million in revenue.7
Kilowatt hour sales to the ultimate customer is the statistic most often used to measure electric utility output. In 1992, State and local systems sold 395,387 million kilowatt hours to ultimate customers. The 10 largest systems accounted for about 30 percent of the kilowatt sales and the 20 largest for about 40 percent. The largest 483
systems sold almost 90 percent of kilowatt sales according to the U.S. Energy Information Administration (EIA).8
The discussion and calculations that follow focus on State and local government
electric utilities whether they generate, transmit, or distribute power. This follows the
Standard Industrial Classification (SIC) system which is used by BLS and most of the
Enterprise Service—34
rest of the statistical community. The SIC assigns all “establishments engaged in the
generation, transmission and/or distribution of electric energy for sale” to SIC code
4911.9 The focus is on the 50 States and the District of Columbia; utilities in the trust
territories are excluded.
Research and statistics abound on electric utility operations. This stems from the
public’s past interest in utility regulation and rate setting, the great debates of the 1930s
over public vs. private power, the recent interest in the safety of nuclear power, and the
effect of acid rain, all issues that lend themselves to economic analysis. Universities,
private consulting firms, utilities, and government regulators routinely study the industry.
Table 9. Finances of State and local government electric utilities by type of government, fiscal year 1992
(millions of dollars)
Category
Total
State
Revenue ....................
Expenditures .............
Capital .....................
Interest on debt .......
Current operations ..
Compensation ......
Other ....................
$30,999
31,983
3,950
5,293
22,739
3,800
18,939
$2,258
2,532
408
571
1,552
341
1,211
Special
County Municipality Township District
$132
144
27
13
104
10
94
$18,232
18,023
2,168
1,353
14,502
2,190
12,312
$511
487
11
3
473
43
430
$9,868
10,797
1,335
3,353
6,108
1,217
4,891
NOTE: “Compensation” estimated by multiplying Bureau of Census October pay times 12.
Standard benefits were derived for general government and applied to salary and wage estimates
to reach total compensation. “Other” is a residual—current operations minus compensation.
SOURCE: U.S. Bureau of the Census, Government Finances: 1991-92, Series GF/92-5. U.S.
Government Printing Office, Washington, DC (1996) and U.S. Bureau of the Census, Public Employment: 1992, Series GE/92-1. U. S. Government Printing Office, Washington, DC (1994).
The EIA, the American Public Power Asociation (APPA), the Bureau of the Census,
investment firms, and individual utilities all publish data on municipal electric power.
There are statistics on the number of customers, kilowatt hour sales, revenues, number
of generating stations, miles of transmission lines, plant cost, and allowances for depreciation and amortization. Data used to construct the State and local government productivity indexes are reviewed in the following sections.
Outputs
The output measure used most often to assess electric utility output is kilowatt hours
sold, and is the measure used in this bulletin. Other measures are the number of customers, kilowatt hours generated, percent of capacity used, generator capacity, dollar sales,
and net profit. The relative strengths and weaknesses of the different measures are
seldom discussed in the literature.
William Iulo, one of the few researchers who has examined the different measures,
offers four reasons for using kilowatt hours:
•
•
•
•
The measure is familiar to industry and the public, and has long been used by
both.
Data are readily available. All utilities collect, keep, and report statistics on
kilowatt hours sold.
The kilowatt hour is a standard physical unit which is not affected by price
changes.
Kilowatt hours are a rough indicator of the industry’s ability to produce electric
energy.
The only argument Iulo offers against the use of the kilowatt hour is that production
costs per kilowatt hour are not always similar.10
Enterprise Service—35
Chapter 2 lists the criteria used in this study to select State and local government
output measures. The kilowatt hour satisfies the four essential criteria and three of the
four optional criteria. The only one not met is the one noted by Iulo: Kilowatt hours are
not always proportional to the cost of producing and delivering the service to the residential, commercial, and industrial consumer.
Production costs vary by class of service. Capital requirements to construct distribution systems for industrial users are normally less than those required to service residential customers per kilowatt hour delivered. Similarly, the labor required to maintain
and service industrial and commercial distribution is likely to be less than that required
for residential service.
Weighting output. Because production costs vary, differentiating or segmenting output
to account for the different classes of service is common practice. BLS’ private sector
electric power productivity calculations currently use seven basic weighted aggregates—
residential, commercial, industrial, public street and highway lighting, other public,
railroads and railways, and interdepartmental.11 In the past, fewer different outputs
among the seven have been used, depending on data availability.
The weights, themselves, should reflect the relative unit cost of producing the service as discussed in chapter 2. However, these data are not always available. When
they are not, unit labor, unit price, or unit revenue are often used as substitutes. In the
case of electric power it is common practice to use unit revenue or average price per
kilowatt hour for each class of service (revenue divided by KWH’s sold) as the weight.
Iulo has shown that there is a good relationship between unit cost and unit revenue for
electric utilities, and BLS uses unit revenue as weights in its investor-owned and cooperative utility calculations.12 This same procedure is used for State and local government electric utility output measurement.13
Utility weights for State and local government utilities have been calculated for 1967,
1972, 1977, 1982, and 1987 (table 10). For 1967, the weights were calculated for
residential, commercial and industrial, and other. Starting in 1972, additional data made
it possible to divide the commercial/industrial field. These are absolute costs. Relative
weights were used to compute the index.
Table 10. State and local government electric utility costs per kilowatt hour by
class of service, selected years
Dollars per kwh
Class of service
Residential .......................
Commercial/Industrial ......
Commercial ....................
Industrial ........................
Other ................................
1967
1972
1977
1982
1987
$0.015
.011
—
—
.015
$0.016
—
.017
.010
.017
$0.029
—
.033
.020
.034
$0.048
—
.051
.033
.055
$0.059
—
.061
.044
.116
NOTE: The large increase in the 1987 “other” weight reflects a large drop in the sales of kwh in
that year. The impact of this change on the overall index is small because “other” accounts for less
than 6 percent of total sales. Dash indicates data not computed.
SOURCE: 1987 computed from statistics provided by the American Public Power Association;
all other years computed from Statistics of Publicly Owned Electric Utilities in the United States,
selected issues (U.S. Department of Energy, Energy Information Administration).
Quality of service. The quality of service is not an issue for most electric power productivity researchers. Whether this is due to conceptual difficulties, data problems, a
feeling that quality is an unimportant issue, or a combination of factors, is not known.
Researchers who have studied the electric power quality issues have singled out the
following as important:
Enterprise Service—36
• Reliability. This factor concerns the number, length, and duration of supply inter•
•
ruptions. Factors such as weather, disaster, lack of equipment, or lack of fuel may
cause interruptions. Building redundancy into the system increases reliability.
Voltage. Lack of proper equipment or insufficient generating capacity may cause
voltage fluctuations which result in damage or malfunction of user equipment.
Installing additional equipment can control voltage fluctuations.
Aesthetics. The aesthetic factor most often discussed is placement of utility lines—
above vs. below ground. Placing utilities below ground increases initial costs. Its
impact on operating costs is open to debate.
Adjustments for quality have not been attempted for State and local government
electric power utility service.
Generation vs. sales. Because State and local government utilities are not a closed
system, a problem may arise in calculating their productivity when they generate and
sell power to non-State and non-local utilities, or alternatively purchase and distribute
power that other utilities generate. North Platte, Nebraska, for example, sold 198.9
million kilowatts in 1992 to ultimate consumers but generated no electricity itself. The
New York State Power Authority, on the other hand, generated 28.6 billion kilowatt
hours in 1992 but sold only 13.2 billion to ultimate consumers; it sold the remaining
kilowatt hours to other utilities, private and government.14
There should be no problem in calculating productivity trends so long as the overall
ratio between generation and sales to the ultimate consumer remains roughly the same
for all State and local government utilities. In the case of State and local government
utilities, the ratio was roughly the same in 1992 as it was in 1967. That is, the utilities,
as a group, generated about 70 to 75 percent of what they sold to the final consumer.
This ratio has fluctuated from year-to-year as new generating capacity came on line or
some large unit was taken off line.15
Even when the ratio of State and local government generating capacity vis-a-vis
sales to ultimate customers shifts (such as in the 1970s when the New York State Power
Authority brought major, new generating plants on line), the impact on average labor
productivity is limited. This reflects the small role that labor plays in electric power
generation. Labor productivity with, and without, the incremental change in employees
working in generation seems to have little effect on overall average labor productivity
calculations.
Statistics. There are two basic sources of data on kilowatt hours sold to ultimate consumers by State and local government utilities: (1) the individual utilities and (2) the
EIA. The EIA collects and publishes summary information annually on all State and
local government utilities (2,017 in 1992) and detailed information on the larger ones,
i.e., those that sold more than 120,000 megawatt hours of electricity; there were 483
large utilities in 1992). In past years, APPA has collected some output statistics from a
sample of its members, but the Association currently relies on EIA for these data.
The EIA statistical reports include, in addition to summary data, details for individual utilities on the number of customers, kilowatt hour sales, revenues, production
expenses, assets, liabilities, profit and loss, generating capacity, number of miles of
transmission lines, and numerous other statistics. Kilowatt hour sales are divided by
class of customer.
The EIA and its predecessor organizations, particularly the Federal Power Commission, have published statistics on publicly-owned utilities since 1946. However, reporting requirements and tabulation procedures were modified a number of times so that
year-to-year summary comparisons are difficult, and can be misleading. Consistent
data, including summary data, are available since 1985. The post-1985 data are far
superior to the pre-1985 data for purposes of current analysis.
Output indexes. In 1992, State and local government utilities sold 395,386 million
Enterprise Service—37
kilowatts of power to ultimate customers, the primary consumers being residential and
industrial. The breakdown by type of customer was: 34.7 percent residential, 25.7
percent commercial, 33.8 percent industrial, and 5.8 percent other.16
Output of State and local government kilowatt sales grew steadily from 1967 to
1992. The average annual rate of growth of weighted output for the entire period was
3.6 percent. In only 2 years out of the 25, was there a drop in the sales to final customers. The fastest growth in output was recorded in the latter part of the 1960s and the
early 1970s. Between 1967-72, the average annual rate of growth was 7.0 percent. The
oil embargo of 1973, with the ensuing price increases and burgeoning energy conservation programs, resulted in a decided slowdown in growth of kilowatt hour sales of electricity. From 1972-92, the rate of growth has been fairly constant, averaging 2.7 percent annually (table 11).
Table 11. Unweighted and weighted State and local government electric
utility output indexes, 1967-92
(1967 = 100)
Year
Unweighted
index
Weighted
index
1967 ......................................................
1968 ......................................................
1969 ......................................................
1970 ......................................................
1971 ......................................................
1972 ......................................................
1973 ......................................................
1974 ......................................................
1975 ......................................................
1976 ......................................................
1977 ......................................................
1978 ......................................................
1979 ......................................................
1980 ......................................................
1981 ......................................................
1982 ......................................................
1983 ......................................................
1984 ......................................................
1985 ......................................................
1986 ......................................................
1987 ......................................................
1988 ......................................................
1989 ......................................................
1990 ......................................................
1991 ......................................................
1992 ......................................................
100.0
108.2
118.7
128.1
137.3
139.9
143.7
145.5
145.7
154.0
163.8
172.2
181.5
184.3
187.3
181.4
185.3
191.1
196.8
201.9
209.3
219.1
223.8
232.0
236.5
237.7
100.0
108.4
118.9
128.5
137.9
140.3
144.1
145.3
145.7
154.1
164.1
172.6
182.0
184.9
188.0
182.2
186.2
192.2
198.1
203.3
211.1
222.3
226.0
235.2
239.8
239.4
Average annual
rate of change:
1967-92 .................................................
3.5
3.6
1972-92 .................................................
1977-92 .................................................
1982-92 .................................................
1987-92 .................................................
2.7
2.5
2.7
2.6
2.7
2.6
2.8
2.6
The data in table 11 reflect kilowatt sales to ultimate customers weighted by revenue
weights categorized by class of service. A simple unweighted index of kilowatt sales to
ultimate customers was also calculated. Both indexes cover 1967-92.
Comparison of the two indexes through time shows little difference. The average
annual rate of growth for the unweighted index, 3.5 percent, contrasts to 3.6 percent for
Enterprise Service—38
the weighted index (table 11), This similarity in trends is due to the stability of the
customer mix. In 1967, residential service accounted for 37 percent of total sales to
final customers; in 1992 the statistic was 35 percent. Later calculations focus on the
weighted index because, conceptually, it is the preferred index.
Two basic sources provided the data to construct the two output series. For 1967-85,
APPA Directory data were used with adjustments.17 For 1985-92, EIA data, as tabulated by APPA, were used.18
Labor inputs
Two labor measures were used in calculating State and local government labor productivity in chapter 2. They are the number of employees and the number of full-timeequivalent employees. An index of the number of full-time-equivalent employees, as
discussed above, should be equivalent to an hours index.
Sources of data. There are five principal sources of employment data on State/local
government-owned power systems: (1) The individual public power systems, (2) the
American Public Power Association, (3) the Energy Information Administration, (4)
the Bureau of Labor Statistics, and (5) the Bureau of the Census. This bulletin uses data
from the Bureau of the Census because they are the only source of State/local government-owned power utility data available for the entire period under review.
Although the Bureau of the Census publishes figures on total employment and fulltime-equivalent employment, there are several potential problems in using these data to
measure electric utility labor productivity. First, Census included only local employees
in its electric utility series prior to 1980. Because output statistics start in 1967 and
include both State and local kilowatt hours, the labor input series should include both
State and local government employment. The 1980 census figures showed 3,000 State
power employees; other data suggest that the figure was closer to 4,000. For trend
determinations, the relative change is often as important as the absolute change. Examination of the large public power utilities operated by State employees show that
they have grown at a much more rapid rate than the local government utilities. To
reflect this increase, this bulletin estimates State data using information provided by the
three largest State utilities. These utilities accounted for about 95 percent of State
electric utility employment in 1980.
A second potential problem is the aggregate nature of the Census statistics. For
example, it is not known how many force account (construction) employees are included in the totals. Force account workers should be excluded from the statistics.
Also, some cities operate multiple utilities, e.g., gas, water, and sewerage, and shift
their employees to the services in which they are needed. How many employees work
in these multiple-service areas is also not known. Overhead personnel are a special
case for the multiple-service utility. Discussions with utility personnel and Census statisticians suggest that these numbers are relatively small, and rough calculations suggest
that they are relatively stable through time and do not affect the labor indexes.
Another potential problem, but probably not for electric power, is that Census employment statistics are collected for only the month of October each year. These statistics do not capture seasonal employment. However, discussions with electric power
officials suggest that seasonality does not substantially affect their employment. For
trend calculations the October figure is probably sufficiently accurate for electric power
employment calculations.
Employment Statistics. There were 85,210 State and local government electric power
employees in 1992 according to the Bureau of Census. Most utilities are small and
have only a few employees, but others, such as the City of Los Angeles and Salt River,
Arizona, have thousands.
Electric power municipal utility employment usually means full-time employment.
Approximately 97 percent of State and local government electric power employees
were full-time employees in 1967 and 1992. The ratio of full-time to total has remained
fairly constant through time. In 1992, there were 83,612 full-time-equivalent employees.19
Enterprise Service—39
Four employment indexes are shown here for State and local government electric
power utility employment (table 12). The first is for total local government electric
power employment, the second is for full-time-equivalent local government electric
power employment. Both sets of data were taken directly from Census Bureau statistics. The third and fourth indexes are for State and local government employees, and
include the adjustment for State employees noted earlier. For 1980-92, the data were
TABLE 12 Four State and local government electric utility labor indexes,
1967-92
(1967 = 100)
Local government
Year
Total number
FTE
of employees employees
State and local
government
Total number
of employees
FTE
employees
1967 ................................
1968 ................................
1969 ................................
100.0
98.3
94.3
100.0
98.2
94.7
100.0
98.3
94.5
100.0
98.3
94.9
1970 ................................
1971 ................................
1972 ................................
1973 ................................
1974 ................................
1975 ................................
1976 ................................
1977 ................................
1978 ................................
1979 ................................
98.3
101.1
100.7
103.7
106.2
105.4
104.3
101.9
106.1
112.1
98.2
101.8
101.8
105.3
107.0
105.3
105.3
101.8
108.8
112.3
98.5
101.3
100.9
103.9
106.5
105.9
105.2
103.3
107.9
113.9
98.4
101.9
101.9
105.5
107.3
105.7
106.1
103.2
110.5
114.1
1980 ................................
1981 ................................
1982 ................................
1983 ................................
1984 ................................
1985 ................................
1986 ................................
1987 ................................
1988 ................................
1989 ................................
115.2
118.1
120.4
120.7
123.0
125.2
128.1
129.3
130.7
127.8
112.3
115.8
116.9
117.4
122.6
124.7
129.8
131.1
131.5
129.7
117.1
120.4
122.1
125.4
127.0
130.3
133.6
134.2
135.6
132.7
114.3
118.5
117.7
122.7
128.5
131.8
135.8
137.3
137.7
135.9
1990 ................................
1991 ................................
1992 ................................
131.1
132.9
133.6
132.6
134.3
135.4
136.0
137.9
140.6
138.8
140.6
144.1
Average annual
rate of change:
1967-92 ...........................
1.2
1.2
1.4
1.5
1967-72 ...........................
1972-77 ...........................
1977-82 ...........................
1982-87 ...........................
1987-92 ...........................
.1
.2
3.4
1.4
.6
.3
.0
2.8
2.3
.6
.2
.5
3.4
1.9
.9
.4
.2
2.7
3.1
1.0
SOURCE: Computed from data taken from Bureau of Census, Public Employment, annual
issues, with adjustments
Enterprise Service—40
taken from Census publications; for 1967-80, State employment was estimated and
added to local government statistics taken from Census data.
The overall average annual rates of growth for the four indexes are quite similar.
The compound rate of growth for 1967-92 for the first two indexes is 1.2 percent. For
the third it is 1.4 percent and for the fourth it is 1.5 percent. In each case there has been
a steady progression in the growth of labor with a slightly greater increase in the 197886 period.
The two FTE indexes seem to grow at a slightly greater rate than the two indexes for
the total number of employees. But the difference is small in each case, and the Bureau
of Census revised its FTE tabulation procedures in 1985, so any conclusions about
these differences should be treated with caution.20
Productivity indexes
Two labor productivity indexes are presented in this section. They use the same
output index, but different labor indexes. One labor index reflects total State and local
government employment and the other, full-time-equivalent employment. Both draw
on the data and investigative approaches presented in the preceding discussion and
cover 1967-92. There is little difference in the rate of change between the two indexes:
the total employment productivity index increased at an average annual rate of 2.2
percent while the FTE index grew at 2.1 percent (table 13).
The rate of change, however, is marked by uneven periods of growth. The early
period, 1967-72, was marked by sizable increases. This was followed by more moderate increases. The latter part of the 1970s and the early and mid-1980s were marked by
decreasing productivity. Since the latter part of the 1980s, modest productivity growth
has returned (table 13).
The large increase in productivity in the early years (1967-72) was driven by large
increases in output. From 1972-92, output growth moderated and average labor productivity was affected by changes in labor. Indeed, the modest increases in output
between 1972 and 1982 were matched by modest increases in labor input. The increase
in output per FTE from 1987-92 was a function of declining, though still positive, rates
of growth in labor input or employment.
Government-private
comparisons
Comparison of State and local government and private (investor and cooperative)
utility labor productivity movements show remarkably similar trends over the long term.
Between 1967 and 1992, the average annual rate of increase was 2.1 percent for State
and local government and 2.3 percent for the private utilities; for 1967-90 the rates
were identical, 2.3 percent (table 14).
These indexes diverge at several points during the period, however. In the early
years, e.g., 1967-72, State and local government productivity index grew at a more
rapid rate than did the private index. The opposite was the case in the 1990s with
private utilities growing more rapidly. Both indexes reflect declines in average labor
productivity in the late 1970s and early 1980s.
Examination of electric power output shows that the private utility kilowatt hour
sales to ultimate customers has grown at a slightly greater rate than it has for State and
local government utilities. The long-term figures (1967-92) are 4.0 percent and 3.6
percent, respectively. In the 25 measured years, output dropped in 3 years for the private utilities and 2 years for State and local government utilities. Output dipped slightly
for both in 1982 and 1992, the first, apparently, due to the recession and the second,
apparently, due to mild weather.21
The labor indexes, FTE in the case of State and local government and hours for
private workers, have also increased at a slightly greater rate for private utilities from
1967 to 1992, 1.7 percent versus 1.5 percent for State and local utilities. However, a
more interesting divergence is the drop in labor hours for the private utilities from
1986-92 while the state and local government utilities continue to add FTE employees,
albeit at a more modest rate than in the past.22
Enterprise Service—41
Table 13. Indexes for State and local government electric utilities: Output, FTE
employee, total employee, output per FTE employee, and output per
employee, 1967-92
(1967 = 100)
Productivity
Year
Output
FTE
employee
Total
Output per
employee FTE employee
1967 ...................
1968 ...................
1969 ...................
100.0
108.4
118.9
100.0
98.3
94.9
100.0
98.3
94.5
100.0
110.3
125.3
100.0
110.2
125.9
1970 ...................
1971 ...................
1972 ...................
1973 ...................
1974 ...................
1975 ...................
1976 ...................
1977 ...................
1978 ...................
1979 ...................
128.5
137.9
140.3
144.1
145.3
145.7
154.1
164.1
172.6
182.0
98.4
101.9
101.9
105.5
107.3
105.7
106.1
103.2
110.5
114.1
98.5
101.3
100.9
103.9
106.5
105.9
105.2
103.3
107.9
113.9
130.6
135.3
137.7
136.6
135.4
137.8
145.3
159.0
156.1
159.5
130.5
136.1
139.1
138.6
136.4
137.6
146.5
158.8
159.9
159.8
1980 ...................
1981 ...................
1982 ...................
1983 ...................
1984 ...................
1985 ...................
1986 ...................
1987 ...................
1988 ...................
1989 ...................
184.9
188.0
182.2
186.2
192.2
198.1
203.3
211.1
222.3
226.0
114.3
118.5
117.7
122.7
128.5
131.8
135.8
137.3
137.7
135.9
117.1
120.4
122.1
125.4
127.0
130.3
133.6
134.2
135.6
132.7
161.7
158.7
154.8
151.8
149.6
150.4
149.7
153.8
161.4
166.2
157.9
156.1
149.3
148.5
151.3
152.0
152.1
157.3
164.0
170.3
1990 ...................
1991 ...................
1992 ...................
235.2
239.8
239.4
138.8
140.6
144.1
136.0
137.9
140.6
169.5
170.5
166.2
172.9
173.9
170.3
Average annual
rate of change:
1967-92 ..............
3.6
1.5
1.4
2.1
2.2
1967-72 ..............
1972-77 ..............
1977-82 ..............
1982-87 ..............
1987-92 ..............
7.0
3.2
2.1
3.0
2.6
.4
.2
2.7
3.1
1.0
.2
.5
3.4
1.9
.9
6.6
2.9
-.5
-.1
1.6
6.8
2.7
-1.2
1.1
1.6
SOURCE: Output, table 11; employees, table 12
Enterprise Service—42
Output per
employee
Table 14. Comparison of State and local government and investor and cooperative-owned electric utility productivity indexes, 1967-92
(1967 = 100)
State and local
government
Investor and
cooperative
1967 ............................................
1968 ............................................
1969 ............................................
100.0
110.3
125.3
100.0
107.6
114.0
1970 ............................................
1971 ............................................
1972 ............................................
1973 ............................................
1974 ............................................
1975 ............................................
1976 ............................................
1977 ............................................
1978 ............................................
1979 ............................................
130.6
135.3
137.7
136.6
135.4
137.8
145.3
159.0
156.1
159.5
118.2
124.8
131.5
135.8
133.6
142.4
146.8
153.7
148.6
146.6
1980 ............................................
1981 ............................................
1982 ............................................
1983 ............................................
1984 ............................................
1985 ............................................
1986 ............................................
1987 ............................................
1988 ............................................
1989 ............................................
161.7
158.7
154.8
151.8
149.6
150.4
149.7
153.8
161.4
166.2
144.3
143.1
137.3
139.7
145.1
143.5
147.1
154.3
161.9
166.2
1990 ............................................
1991 ............................................
1992 ............................................
169.5
170.5
166.2
169.9
175.0
176.4
Compound rate:
1967-92 .......................................
2.1
2.3
1967-72 .......................................
1972-77 .......................................
1977-82 .......................................
1982-87 .......................................
1987-92 .......................................
6.6
2.9
-.5
-.1
1.6
5.6
3.2
-2.2
2.4
2.7
Year
SOURCE: State and local government, table 13; Investor and cooperative-owned, table 149,
Productivity Measures for Selected Industries and Government Services, (Bureau of Labor Statistics, Bulletin 2440, 1994), p. 81.
Natural Gas
Natural gas utilities, like electric power, are operated primarily by the private sector,
produce a measurable set of outputs, and report frequently on their operations. Furthermore, private natural gas utility operations are often measured and labor productivity
indexes routinely calculated. However, unlike electric power, natural gas local government productivity indexes have not been calculated.
Institutional setting
Private investor-owned companies dominate natural gas sales in the United States.
Almost 94 percent of the sales to final customers are by private, for-profit companies
and 93 percent of all gas employees are private sector employees.23
There are approximately 1,400 local gas distribution companies (LDC) in the United
States. These companies purchase gas from pipeline companies and resell it to residenEnterprise Service—43
tial, commercial, and industrial customers. The U.S. natural gas industry also produces
and transmits gas. However, production and transmission are not a concern of this
section.
This section focuses on distribution, or sales to the ultimate customers. It excludes
production and transmission. This differs from the approach used for electric power
utilities which included generation and transmission to the utility as well as distribution
to the customer. This distinction follows the nomenclature used in the Standard Industrial Classification Manual which assigns natural gas distribution sales to SIC 4924,
which includes all “establishments engaged in the distribution of natural gas for sale.”24
Other SIC codes are used for production and transmission.
The natural gas industry served about 55 million customers in 1991. Ninety-two
percent were residential and 8 percent were commercial. Industrial and other comprised less than 1 percent. Dollar revenue sales were more evenly divided. Residential
made up about 58 percent, commercial 24 percent, industrial 17 percent, and other 1
percent. Total gross revenue was $44.6 billion in 1991.25
In local government, there were 948 jurisdictions in 1992 providing natural gas service. The breakdown by type of government was: Municipal 849, county 17, townships
16, and special districts 66.26
Local government gas utilities are concentrated in Indiana, Tennessee, and Texas,
and account for about 45 percent of all local gas employment. Fourteen States and the
District of Columbia report no local government natural gas utility operations. 27
Gross revenue of local government systems was just over $3 billion in 1992. Expenditures were $3.1 billion, of which operations accounted for 83 percent. Fuel, materials, supplies and purchased gas accounts for most operating expenditures.28 Labor
compensation accounted for only about 14 percent of operating expenditures (table
15).29
Table 15. Finances of local government natural gas utilities by type of government, fiscal year 1992
(millions of dollars)
Category
Total
County
Revenue ....................
Expenditures .............
Capital .....................
Interest on debt .......
Current operations ..
Compensation .......
Other .....................
$3,034
3,058
432
92
2,533
364
2,169
$15
30
15
—
14
4
10
Municipality
$2,321
2,177
198
71
1,908
259
1,649
Township
Special
district
$9
9
—
—
9
1
8
$684
839
219
21
599
99
500
NOTE: “Compensation” estimated by multiplying Bureau of Census October pay times 12.
Standard benefits were derived for general government and applied to salary and wage estimates
to reach total compensation. “Other” is a residual—current operations minus compensation.
SOURCE: U.S. Bureau of the Census, Government Finances: 1991-92, Series GF/92-5. U.S.
Government Printing Office, Washington, DC (1996) and U.S. Bureau of the Census, Public Employment: 1992, Series GE/92-1. U. S. Government Printing Office, Washington, DC (1994).
There are generally reliable statistics on local government gas operations. The EIA,
the America Gas Association (AGA), the Bureau of Census, investment firms, and individual gas utilities all publish data on natural gas operations.30 There are statistics on
the number of customers, dollar sales, amount of gas sold, miles of transmission mains,
plant cost, allowances for depreciation, return on investment, and number of employees.
Enterprise Service—44
Outputs
The two measures that seem to be used most often when discussing and calculating
gas utility output are the number of British thermal units(BTU’s) sold and the number
of cubic feet sold.31 Other output measures are the number of customers, dollar sales,
gas transported, gas produced, and net profit. No discussion was found in the research
literature of the relative strengths and weaknesses of the different measures. Nor was
there any discussion of which measure was preferred for labor productivity measurement.
Chapter 2 listed the criteria used in this paper to select State and local government
output measures. The BTU and cubic foot measures satisfy both the essential and
optional criteria. Although both the cubic foot and the BTU are satisfactory measures
of utility output, the BTU measure is used in this bulletin. It is also the indicator used in
BLS calculations of private sector output measures.
Local government natural gas utilities sold 741 trillion BTU’s in 1974, the first year
for which data are available. By 1992, sales had dropped to 616 trillion, a decrease of
17 percent. The average annual decrease during this period is 1.0 percent. This drop
follows rapidly increasing prices which apparently fostered greater efficiencies and
switches to alternative fuels (table 16).
The breakdown of delivery by type of consumer between 1974 and 1992 shows a
varied picture. Specifically, residential, industrial, and other deliveries have decreased
6 percent, 40 percent, and 63 percent, respectively, while commercial deliveries have
increased by 23 percent. All statistics are for 1974-92.
Table 16. Natural gas delivery by local government utilities by class of service,
1974-92
(trillions of BTU’s)
Year
Residential Commercial
Industrial
Other
Total
1974 .......................
1975 .......................
1976 .......................
1977 .......................
1978 .......................
1979 .......................
300
304
302
301
319
308
122
127
127
128
131
129
289
231
204
176
187
236
30
27
25
25
27
28
741
689
658
630
664
701
1980 .......................
1981 .......................
1982 .......................
1983 .......................
1984 .......................
1985 .......................
1986 .......................
1987 .......................
1988 .......................
1989 .......................
293
281
285
271
286
266
264
265
280
282
126
128
134
126
141
137
136
138
140
144
241
244
224
212
213
182
145
152
168
165
26
19
20
25
25
20
16
18
17
17
686
672
663
634
665
605
561
573
605
608
1990 .......................
1991 .......................
1992 .......................
262
271
283
138
141
150
165
168
172
15
11
11
580
591
616
SOURCE: Gas Facts, annual issues
Natural gas end users usually pay different unit prices depending on the type of
use—i.e., residential, commercial, industrial, or other—and type of service received—
i.e., firm or interruptible supply. The price differences usually reflect the differences in
Enterprise Service—45
the cost of providing service. Residential customers, for example, are usually widely
dispersed, use relatively small amounts of gas, but need peak amounts on extremely
cold days. Service to them is more costly per unit than to industrial users who have
relatively stable demand and use gas in large quantities.
Weighting output. When production costs vary by type of service it is important to
differentiate the types of service for productivity measurement, and weight them accordingly. BLS’ private sector natural gas output calculations use four basic weighted
aggregates: Residential, commercial, industrial, and other.32 The same breakdown is
used for local government natural gas output measurement. Each output is weighted by
its unit price in the base year and the weighted segments are combined to create the
overall index.33
The weights should reflect the unit cost or relative cost of producing the service as
discussed in chapter 2. However, these data are not always available and unit labor,
unit price, or unit revenue are used as substitutes. For natural gas, unit revenue or
average price per therm for each class of service (revenue divided by therms sold) is
used as the weight in the private sector measure, and that procedure is used here. Relative weights are used in the actual calculations.
The costs per therm by class of service for local government natural gas utilities are
presented in table 17. These weights were taken from EIA calculations for the private
sector because data were not available with which to calculate local government weights.
It is not known whether the relative prices charged by local government utilities are the
same as those charged by the private utilities, however, it seems likely that the two
would move in concert.
The price weights show rapidly increasing unit prices for residential sales. In 1974,
the unit price was $1.43 per therm. By 1987, it had reached $5.54. The unit price rose
for the other three classes of service over the 1974-82 period but decreased in 1987.
According to industry personnel, this is a reflection of the deregulation of natural gas,
the subsequent shifting of costs and prices and, in particular, agreement by buyers to
have their gas supply temporarily interrupted.
Table 17. Local government natural gas utility costs per therm by class of
service, selected years
Class of service
Residential .......................
Commercial ......................
Industrial ..........................
Other ................................
Dollars per therm
1974
1977
1982
1987
$1.43
1.07
0.67
0.51
$2.35
2.04
1.50
1.32
$5.17
4.82
3.87
3.48
$5.54
4.77
2.94
2.32
NOTE: Residential: Service to customers for domestic purposes including single, multifamily,
and mobile homes. Commercial: Service to customers engaged in wholesale or retail trade. Industrial: Service to customers engaged primarily in extraction of raw materials or changing unfinished materials into another form. Other: Service to local, State, or Federal Government; excludes
enterprise service sales.
Output index. The weighted output index shows an average annual decrease of 0.7
percent between 1974 and 1992 (table 18). The unweighted index, as noted above,
shows an average annual decrease of 1.0 percent. Where electric power indexes show
no difference in the average annual change between the weighted and unweighted (2.8
percent) indexes for 1974-92, the natural gas weighted output index reflects the shift to
Enterprise Service—46
Table 18. Unweighted and weighted local government
natural gas output indexes, 1974-92
(1974=100)
Unweighted
output
index
Weighted
output
index
1974 ...........................
1975 ...........................
1976 ...........................
1977 ...........................
1978 ...........................
1979 ...........................
100.0
93.0
88.8
85.0
89.6
94.6
100.0
96.2
93.3
90.8
95.7
98.9
1980 ...........................
1981 ...........................
1982 ...........................
1983 ...........................
1984 ...........................
1985 ...........................
1986 ...........................
1987 ...........................
1988 ...........................
1989 ...........................
92.6
90.7
89.5
85.6
89.7
81.6
75.7
77.3
81.6
82.1
96.3
94.2
93.7
89.4
94.1
86.2
80.9
82.4
86.7
87.4
1990 ...........................
1991 ...........................
1992 ...........................
78.3
79.8
83.1
82.8
84.8
88.7
Average annual
rate of change:
1974-92 ......................
-1.0
-0.7
1974-77 ......................
1977-82 ......................
1982-87 ......................
1987-92 ......................
-5.3
1.0
-2.9
1.5
-3.2
.6
-2.5
1.5
Year
residential and commercial from industrial and other sales. The weighted output index
is used for the calculations presented and discussed later in this chapter.
Labor inputs
There were 10,561 local government natural gas employees in 1992 according to the
Bureau of Census. Most local government utilities are small and have only a few employees.
Gas utility employment, like electric power, means full-time employment. Approximately 96 percent of all local government electric power employees were full-time
employees in 1992, and the ratio of full-time to total has remained fairly constant through
time.
As discussed in chapter 2, a full-time-equivalent employee index is comparable to
an hours index. There were 10,392 FTE employees in 1992.34
Because most gas employees are full-time employees, and the ratio between full
time and part time has remained fairly constant throughout time, the only labor index
constructed and presented here is an FTE series. The data to construct this index are
Census Bureau data. The index shows an average annual rate of growth from 1974-92
of 0.9 percent (table 19).
Enterprise Service—47
Table 19. Indexes of output, FTE employment, and output per FTE
for local government natural gas utilities, 1974-92
(1974=100)
Year
Output
FTE
employment
Output
per FTE
1974 .................................
1975 .................................
1976 .................................
1977 .................................
1978 .................................
1979 .................................
100.0
96.2
93.3
90.8
95.7
98.9
100.0
99.1
98.3
97.4
101.5
105.0
100.0
97.0
95.0
93.2
94.3
94.2
1980 .................................
1981 .................................
1982 .................................
1983 .................................
1984 .................................
1985 .................................
1986 .................................
1987 .................................
1988 .................................
1989 .................................
96.3
94.2
93.7
89.4
94.1
86.2
80.9
82.4
86.7
87.4
96.1
95.6
103.1
99.9
100.7
104.8
109.1
104.9
111.5
114.7
100.2
98.6
90.9
89.5
93.5
82.2
74.1
78.5
77.7
76.2
1990 .................................
1991 .................................
1992 .................................
82.8
84.8
88.7
114.0
113.4
118.0
72.7
74.8
75.1
Average annual
rates of change:
1974-92 ............................
-0.7
0.9
-1.6
1974-77 ............................
1977-82 ............................
1982-87 ............................
1987-92 ............................
-3.2
.6
-2.5
1.5
-.9
1.1
.4
2.4
-2.3
-.5
-2.9
-.9
Productivity indexes
The increasing employment index and decreasing output index results in falling labor productivity. The average annual decrease between 1974-92 is 1.6 percent, and
each 5-year period shows decreasing productivity. In 12 of the 18 measured years
productivity declined (table 19).
Government-private
comparisons
Enterprise Service—48
Investor-owned gas utilities also registered decreasing labor productivity over the
past two decades (table 20). Between 1974 and 1992, the average annual change for
investor utilities was -2.2 percent.35 For local government, as noted above, it was -1.6
percent. Between 1974 and 1980 there was little change in either index, but, beginning
in 1980, both indexes decreased. The average annual change between 1980 and 1992
was -3.3 percent for investor-owned utilities and -2.4 percent for local governmentoperated systems.
Both investor and government operations are marked by falling output which is apparently the driving force behind the decreasing labor productivity, at least through
1987. The average annual change between 1974 and 1987 in investor output is -2.1
percent, for government it is -0.7 percent. Investor labor hours remained fairly stable
over the measured period while local government FTE’s, as noted above, increased
slightly.
Table 20. Comparison of local government and investorowned natural gas utility labor productivity indexes, 1974-92
(1974=100)
Index
Water Supply
Year
Local
government
Investorowned
1974 ...............................
1975 ...............................
1976 ...............................
1977 ...............................
1978 ...............................
1979 ...............................
100.0
97.0
95.0
93.2
94.3
94.2
100.0
99.4
101.7
98.1
99.5
101.4
1980 ...............................
1981 ...............................
1982 ...............................
1983 ...............................
1984 ...............................
1985 ...............................
1986 ...............................
1987 ...............................
1988 ...............................
1989 ...............................
100.2
98.6
90.9
89.5
93.5
82.2
74.1
78.5
77.7
76.2
100.1
96.3
87.3
79.7
82.0
80.6
72.7
70.6
74.4
73.0
1990 ...............................
1991 ...............................
1992 ...............................
72.7
74.8
75.1
66.9
66.2
66.9
Average annual
rate of change:
1974-92 ..........................
-1.6
-2.2
1974-80 ..........................
1980-92 ..........................
0
-2.4
0
-3.3
1974-77 ..........................
1977-82 ..........................
1982-87 ..........................
1987-92 ..........................
-2.3
-.5
-2.9
-.9
-.6
-2.3
-4.1
-1.1
In contrast to electric power and gas utility operations, which are mostly private
sector operations, water supply is primarily a government service. Expenditure,
employment, and population-served statistics show that about 85 percent of U.S.
water utility operations are government operations. Water operations, like electric
power and gas, are capital intensive and are supported primarily by fees and charges.
Institutional setting
There are about 57,000 community water systems in the U.S., serving about 240
million people.36 A community system, according to the U.S. Environmental Protection Agency (EPA) definition, is one that serves 25 or more year-round residents or
which has 15 or more connections to permanent residences. Most of the U.S. population, especially the urban population, is served by government-owned and operated
community water system utilities. In 1992, there were about 17,800 government systems.
Large utilities provide most of the drinking water in the United States. The largest 1
percent serve 44 percent of the population, the largest 13 percent serve almost 90 percent. At the other end of the scale, 87 percent of the systems serve almost 10 percent of
the population.37 All statistics are for total community water systems, private and government.
The EPA estimated total drinking water operating costs in 1987 at $11.8 billion.
Government expenditures were estimated at $9.9 billion (84 percent) and private at
$1.8 billion (16 percent). Capital outlays were another $7.1 billion; government utiliEnterprise Service—49
ties spent about $6.0 billion (85 percent) and private $1.1 billion (15 percent).38
Gross revenue of State and local government water supply utilities was about $19.1
billion in 1992. Expenditures were $24.7 billion of which capital and interest was
$10.9 billion and operations were $13.5 billion. Operating expenditures can be divided
between purchased power, supplies, contract services, and the like, which accounted
for 61 percent, and compensation (wages, salaries and fringe benefits) which accounted
for 39 percent (table 21).
Table 21. Finances of State and local government water supply utilities by type
of government, fiscal year 1992
(millions of dollars)
Category
Revenue ....................
Expenditures .............
Capital .....................
Interest on debt .......
Current operations ..
Compensation .......
Other .....................
Total
$19,147
24,738
7,567
3,290
13,521
5,300
8,221
State
County Municipality Town- Special
ship District
$124
209
88
33
88
37
51
$1,424
2,070
794
296
980
403
577
$12,724
14,661
4,098
1,861
8,702
3,497
5,205
$572
687
184
61
442
133
309
$4,304
6,751
2,403
1,039
3,309
1,230
2,079
NOTE: “Compensation” estimated by multiplying Bureau of Census October pay times 12.
Standard benefits were derived for general government and applied to salary and wage estimates
to reach total compensation. “Other” is a residual—current operations minus compensation.
SOURCE: 1992 Census of Governments—Government Finances 1991-1992 and Public Employment 1992
Government-operated water utilities are, for the most part, local government systems. Three States, Massachusetts, Nevada and New Jersey, operate drinking water
utilities, but employ fewer than 1,000 persons in total. Personnel and expenditure totals
suggest that 99 percent of State and local government water operations are local government operations. Most local employees work for municipalities (72 percent); others
work for counties and special districts. There are local and State government employees who work on water planning and environmental issues but they are not considered
here.39
More than 17,800 local government agencies deliver drinking water. All States have
some local government water utilities, and some have many. In 1992, the number of
utilities ranged from 4 in Hawaii to 1,044 in Illinois. The States with the largest number
of local utilities or agencies are Illinois, Texas, and Pennsylvania. Those with the fewest are Hawaii, Rhode Island, and Delaware.40
Some local governments contract with private firms to operate their water utilities.
In 1992, 9 percent were contract and 91 percent were government operated. The focus
of this analysis is government-operated facilities.
Economies of scale mark water utility operations.41 One review noted that largesystem unit costs were about half those of small systems.42 However, another study
found economies even in small- sized systems.43
Economies of scale and the need to improve water treatment are manifested in large
capital investments. Local water utilities spent $7.6 billion in 1992 for capital additions which is 31 percent of all expenditures. By comparison, gas and electric utilities
spent 14 and 12 percent, respectively, on capital investment.44
Labor accounts for about 39 percent of government water utility current operating
expenditures as calculated from table 21. However, several special studies have found
somewhat higher percents spent on compensation. An EPA study of 12 large water
utilities in the 1970s found that labor costs accounted for 42 percent of the utilities’
operating costs.45 In the Cincinnati water system labor accounted for 62 percent of
operating costs in 1973.46 A recent operating budget from the Fairfax County, Virginia
Enterprise Service—50
water utility shows wage and benefits to be 52 percent of the total. The percent compensation, of course, is affected by a number of considerations including the age of the
facility, degree of automation, the purity of the untreated water, the amount of treatment, and even the accounting procedures used.
The water services covered in this section are those included in SIC 4941:
Establishments primarily engaged in distributing water for sale for domestic, commercial, and industrial use. Systems distributing water primarily
for irrigation service are classified in Industry 4971.47
Agencies that monitor water quality, including regulation, research, and planning,
are assigned to SIC 9511, and are not included here. This includes most State agencies
which plan and monitor drinking water supplies.
Quality concerns
Outputs
Probably the greatest challenge and change in drinking water utility operations over
the past two decades is the increase in treatment and testing. Water utilities, particularly
the larger ones, have treated and tested their water for years (the Public Health Service
issued recommended standards for many years). However, it was not until the passage
of the Federal Safe Drinking Water Act of 1974 that the standards were made mandatory. Amendments to the Act in 1986 further tightened the standards. In 1986, 23
contaminants were identified as potentially harmful, but by 1993 the number had reached
84 and the number is expected to grow.48 Just because a contaminant has been identified does not mean that the EPA has issued a standard concerning its testing and treatment. But the identification of the contaminant is the first step along the road.
There are often several ways to remove a contaminant, each with its own efficacy
and cost. Furthermore, the technology changes constantly and its cost continues to
change, hence it is difficult for utilities to estimate the appropriate treatment technology
and resulting cost.49
All community water systems are required to test their drinking water, and most do.
Today, all large utilities, and most small and medium size utilities test and treat their
water. Surveys suggest that between 1975 and 1985 the number of community water
supply systems (private and government) which treat any part of their water increased
from about 60 percent to 97 percent.50
Drinking water treatment and testing costs were estimated at $1.5 billion in 1986.
EPA estimates that the 1986 revised standards will increase annual treatment and testing (abatement and control) cost to $2.5 billion (in 1986 dollars), a 24-percent increase
in utility operating and maintenance cost.51 Other projections show even larger increases.52 These statistics do not include the capital cost associated with treating and
testing water, which is estimated by EPA to be about $1 billion in 1986.53 While these
cost estimates are surrounded by considerable uncertainty, they do highlight the increased costs associated with meeting the EPA water quality standards.
Dozens of measures are used to assess the output of water supply operations. 54 Most
focus on water delivery and sales. Some water utilities produce additional services,
such as recreation programs at reservoirs and water conservation audits, but these programs are a small part of local government water utility services and expenditures. This
section focuses on the supply of drinking water to the community.
Water utility operations are usually divided into four functions or parts—acquisition, treatment, distribution, and overhead. Acquisition, normally a small part of the
cost of water, includes all operations before treatment, such as withdrawal of water
from above or underground rivers, storage, and transportation to the treatment facility.
Treatment includes any purification and testing of the water before distribution. It is an
increasing portion of water utility costs. Distribution is delivery of the water to the
customer. Overhead includes all administrative and customer services required to manage a utility. Overhead and distribution account for the largest portion of the operating
expenditures of most utilities.
Enterprise Service—51
Our examination focuses on final output: the delivery of water to the public. Individual utility functions or process are not examined. Testing and treatment are considered but in the context of how they affect final output. The cost of delivery varies by
type of consumer, and this is considered as well.
Five indicators of water supply output are examined in the following discussion:
Revenue gallons sold, gallons produced, connections, population served, and revenue
or dollar sales (deflated value). These measures seem to be used most often in measuring gross output. The strengths and weaknesses of each candidate measure are briefly
reviewed using the criteria discussed in chapter 2.
Revenue gallons sold. The preferred measure for tracking a water utility’s output, at
least for the purposes of measuring productivity, is the revenue gallon or the gallons of
revenue producing water delivered to the customer. This measure is analogous to kilowatt hours sold, in the case of electric utilities, and British thermal units (Btu) sold, in
the case of natural gas utilities. The revenue gallon is the “basis on which utilities
obtain their operating revenues and provides the real basis for comparing productivity
and costs between systems.”55 This measure is simple, straightforward, repetitive, and
measurable.
This section looks at revenue producing water, that is, the actual water sold or delivered to the customer. Water collected, treated and/or pumped, but lost through leaky
water mains, open hydrants and evaporation, is not counted. Also, by focusing on sales
to the final customer, we remove inter-utility water sales as a potential measurement
problem. The magnitude of resales is not known, but focusing on sales to ultimate users
minimizes the problem of double counting in a national index.
The primary argument against using revenue gallons as the measure of output is the
lack of data. In short, there is no national statistical series of revenue gallons. Indeed,
data have not been collected for even a single year. Furthermore, some water systems
do not even collect statistics on the quantity of revenue water sold. For example, prior
to 1980, New York City, billed by number of meters and, consequently, had no record
of the number of gallons sold to its customers. And the water utilities serving areas
around Denver, Fresno, Reno, Sacramento, Saint Louis, and Schenectady did not meter
most of their residential customers in 1984.56 Even those cities that have a policy of
metering, such as Boston, New Orleans, and Washington, DC, do not meter all their
sales.57 However, most large and medium-size water systems do meter most of their
sales. And metering is becoming more prevalent.
To summarize the revenue producing water measure: It is a utility’s final output, it is
measurable, repetitive, easily understood and a physical measure. The one criterion
where this measure falls short is data availability. While most utilities collect these data
there is apparently no reliable data source from which to compute a national output
index.
Gallons produced. A proxy measure for revenue gallons sold is the number of gallons
produced, such as the amount of water treated or pumped. Such indicators are measurable, physical, and available from most utilities. Utilities may calculate these statistics
when they lack data on revenue gallons sold.
Nevertheless, there are several arguments against using gallons produced as a measure of output. First, although it is a measure of the work performed, it is not a measure
of final output, that is, the number of gallons sold. As noted above, some of the water
pumped or treated is never delivered to utilities’ customers; it is used to flush streets,
fight fires, or is lost through system leakage. Second, this measure means different
things to different people (e.g., gallons treated or gallons pumped) and is open to different interpretation by the utilities. It is not easy to calculate a national series when
definitions are murky. Third, in the case of treated water, more and more utilities are
treating their water, and counting gallons treated may introduce an upward bias in the
output index. Fourth, the national statistics that exist in this area are incomplete and/or
subject to question.
Enterprise Service—52
Connections. The number of connections is simply the number of drinking water hookups. Another way of measuring the number of connections is the number of meters, but
as discussed earlier, not every jurisdiction meters all its water.
This measure can be viewed in one of two ways. On the one hand, it is a reflection
of the work of a utility, that is, the number of connections installed and serviced, and the
number of meters read. The measure captures a part of a water utility’s work. In this
sense it is a work measure. On the other hand, it can be viewed as a surrogate measure
for the revenue water delivered, and it is in this context that the number of connections
is examined in this bulletin.
When this measure is used as a proxy measure, an assumption is made that water use
per connection remains constant through time, or at least, that the cost of delivering
water to each connection remains constant. Some research suggests that this was the
case nationally between 1960-85 as water use per capita increased but the number of
persons per household decreased. To quote one expert, “it is probably best to conclude
that domestic water use per household was essentially the same during 1980 as during
1960.”58
There are problems with this indicator as a measure of output, however. First, most
water utilities serve non-residential as well as residential customers. If a community
has a large number of nonresidential users, or if the ratio between the two changes
through time, the number of connections may not be a good measure of the water sold.
One study estimated that residential units made up 90 percent of the billings of U. S.
water systems but accounted for only 60 percent of the water delivered. If the proportion of residential and nonresidential connections, and the use per connection remains
constant through time, the measure might be satisfactory for trend determinations. But
these ratios are not known.
Although the connection indicator is measurable, repetitive, easily understood, and
a physical measure, it does not reflect the final output of water utilities. Also, there is a
question of the availability and accuracy of national data required to calculate this measure. In short, it is a poor measure of water utility output.
Population served. As with the number of connections, the population served is a
surrogate measure for the number of revenue gallons sold. The basis for this measure is
expert opinion and studies suggesting that over the long run there is a good correlation
between the water consumed and the number of people served. Population figures are
often used to plan future community water needs. Most water utilities keep this statistic
or can readily produce it (number of residential connections times average household
population or simply a count of a jurisdiction’s population). Also, most national surveys of water utilities collect and use these data. This indicator is recommended by
some experts because of its availability and accuracy.
Some research suggests that this indicator is not a very good measure of output,
particularly over the short run. Variations in temperature and rainfall can dramatically
affect water usage. Also, if there is a shift between non-residential and residential water
use, then a simple count of the population is likely to be a misleading proxy for revenue
producing water.
Although this indicator is measurable, repetitive, physical, accurate, and easily understood it does not reflect unit labor requirements spent in the production of the service. It may be satisfactory for planning but it is not very useful for annual productivity
computations.
Revenue or dollar sales (deflated value). Deflated revenue is an indirect way of calculating output. That is, by removing price change one can determine the quantity sold
(price x quantity = revenue). The deflated value approach is often used to measure
private sector output when revenue and price indexes are available. But the approach is
rarely used in the public sector where services are not normally sold, and revenue figures are lacking. However, government enterprise services, such as drinking water, are
largely supported through their sales, and in such instances deflated value should be an
Enterprise Service—53
adequate measure for calculating output.
To calculate a deflated revenue index requires good revenue statistics and price deflators. Although the basic output measure suggested here for water utilities is total
revenue, data are sometimes available by type of sale, such as residential, commercial,
industrial, audits, inspections, and hookups. Such detailed output statistics should produce a better measure of output, than a simple total of all revenue, assuming that price
deflators are also available by type of sale. That is, if revenue data are available for
hookups there needs to be a deflator for hookup charges. Likewise, if revenue sales are
divided between residential and non-residential sales then separate deflators are needed
for these two types of sale.
The U.S. Bureau of Census has collected local government water utility revenue
data for decades. A fairly large sample is collected every year with a complete enumeration every 5 years. The data are used in preparing the national income accounts.
However, there is no national drinking water sales deflator. The only deflator that
comes close to meeting the requirement is the BLS CPI sewer-drinking water deflator.
But, there are many problems and questions surrounding the use of the CPI as a deflator
to compute a water output index. The CPI includes sewer as well as water charges, it is
for urban consumers only, and covers private as well as government sales. The primary
advantage of the CPI deflator is that it is available for the entire period for which we
have revenue statistics.
To summarize, deflated value has a number of distinct strengths as a measure of final
output. It reflects final output, is measurable and repetitive, uses existing data, and
should be accurate and comparable. However, it is not easily understood, and it is
directly affected by the accuracy of the deflators.
Product differentiation. Irrespective of whether the output measure is revenue gallons,
connections, or dollar sales, it needs to be differentiated by type of service. This discussion notes that there are several different types of customers for water utility ouput.
Research shows that unit costs can vary dramatically by type of customer. For example,
the cost of delivering a gallon of water to a residence can be twice that of delivering it
to an industrial firm. The cost of maintaining water mains, billing customers, and even
treating the water can be substantially higher for the residential customer. Also, the
customer mix has undergone substantial shifts over the past several decades. In particular, industry has moved to recycle and reclaim much of the water it uses as water rates
have increased, which has reduced its demand for water. These movements should be
reflected in the output measure used for productivity calculations.
Five different types of customers or classes of water—residential, commercial, industrial, public and wholesale—are commonly serviced by water utilities. Water rates
often reflect this division at the local utility, and several sample surveys have collected
this information. However, no national time series is divided by class of service. Data
availability dictates the degree to which outputs by class of service can be differentiated.
Quality adjustments. Another factor that needs to be addressed when calculating water
supply output is water quality, a subject that has assumed increasing importance in
recent years. How quality affects production costs is the focus of this section.
The costs of delivering drinking water can be significantly affected by the amount of
testing and treatment, and most jurisdictions do test and treat their water. Furthermore,
the amount of testing and treatment has increased dramatically in recent years.
As discussed in chapter 2, there are several ways to handle changes in quality. One
approach is to differentiate the service output by its different levels of quality, weight
each, and then combine them to reach an overall index. The problem with this approach
in the case of drinking water, is that dozens of elements are routinely tested, and when
Enterprise Service—54
treatment is required to remove an element, different types of treatment are used. In
short, there are hundreds of possible weight combinations, and while it is possible to
reduce their number it is still a formidable task to find the required data and combine
them into a single index. Clearly, this approach is not feasible for drinking water.
Another approach to quality adjustment, is to adjust the amount of resources (e.g.,
labor) used to improve quality. In the case of drinking water, the resources used to treat
the water might be estimated and then removed. This approach, which would reduce
the resources used to produce the water, would not penalize a utility’s productivity
measure or index for increasing treatment. This general approach is used here but in a
slightly different form.
The output index. This section has reviewed the strengths and weaknesses of the five
candidate output measures, the desire to differentiate the measure, and the need to adjust for changes in water treatment. Unfortunately, the review does not point to an
obvious output measure for water utilities. There are a number of issues and concerns,
but data availability is the primary problem.
In summary, if the goal is to calculate a water utility output measure, a deflated value
must be chosen as the measure of output. It is the only measure for which government
water supply data exist for the entire 1967-92 period. It is the approach used here.
The data for the basic calculations are taken from the annual Census of Governments
finance series. The Census data series starts in the 1950s, but in keeping with the timing
of other State and local government service indexes, the BLS series begins in 1967.
Annual financial series data are used for all years except for 1967, 1972, 1977, 1982,
and 1987, the Census benchmark years, when the benchmark data are substituted for
the annual series data. Census data show that local government water utility revenue
has increased from $2.2 billion in 1967 to $19.0 billion in 1992, and that there have
been increases every year. Much of the increase in revenue, as we have discussed, is a
reflection of unit price increase over the period (table 22).
To remove the price change, the finance series is deflated with the annual CPI-U
sewer-drinking water deflator. The CPI is the only series that comes close to meeting
the deflator requirements. However, it includes sewer charges as well as water prices,
and there is no way to separate the two. It may be that sewer rates have increased at the
same rate as water, and if they have, the CPI should be satisfactory to use as the deflator
for drinking water utility sales. However, this is not known. Indeed, from what is
known, it is likely that sewer charges have risen more rapidly than drinking water charges
over the past 20 years. If this is the case, the use of the CPI to deflate water revenue will
result in an understatement of the real growth in water output.
There are several other questions concerning the use of the CPI as a local government water utility sales deflator. One, it focuses on urban consumers, and while most
water sales are in urban areas some of the largest increases in water prices have originated with the small, rural utilities. These utilities have had to substantially upgrade
their plant and equipment over the past decade, and supposedly raise their rates to a
greater extent than the cities. If this is the case, use of the CPI will result in an overstatement of the growth of output. Two, the deflator focuses on one type of sale, residential.
Commercial, industrial, and wholesale sales are excluded but they are included in the
revenue figures. It is not known whether the price increases at the same rate for each
type of sale; there are no price deflators for these sales. Three, the CPI reflects private
as well as government sales. However, this should not be a problem since most water
sales are by government—about 85 percent—and the research shows no statistically
significant difference between the costs of government and private production. Four,
no adjustments are made in the CPI for changes in water quality. This probably leads to
an understatement in the increase in water output.
Enterprise Service—55
Table 22. Local government water utility revenue series and indexes, 1967-92
Local
government
revenue
(in thousands
of dollars)
CPI
deflator
1967=100
Deflated
revenue
(in thousands
of dollars)
Deflated
revenue
index
1967=100
1967 .............................
1968 .............................
1969 .............................
$2,187
2,313
2,464
100.0
104.6
111.6
$2,187
2,212
2,208
100.0
101.1
101.0
1970 .............................
1971 .............................
1972 .............................
1973 .............................
1974 .............................
1975 .............................
1976 .............................
1977 .............................
1978 .............................
1979 .............................
2,687
2,980
3,171
3,463
3,712
4,142
4,463
4,989
5,505
6,242
120.4
133.3
138.2
146.0
154.7
169.8
188.4
208.8
232.3
243.2
2,233
2,235
2,294
2,372
2,399
2,439
2,369
2,390
2,370
2,567
102.1
102.2
104.9
108.5
109.7
111.5
108.3
109.3
108.4
117.4
1980 .............................
1981 .............................
1982 .............................
1983 .............................
1984 .............................
1985 .............................
1986 .............................
1987 .............................
1988 .............................
1989 .............................
6,756
7,699
8,451
9,528
10,467
11,947
13,202
14,334
15,243
16,678
259.6
290.5
325.3
352.3
375.4
397.9
418.9
441.4
465.6
494.0
2,602
2,650
2,598
2,705
2,788
3,003
3,151
3,247
3,274
3,376
119.0
121.2
118.8
123.7
127.5
137.3
144.1
148.5
149.7
154.4
1990 .............................
1991 .............................
1992 .............................
17,565
17,920
19,023
527.0
565.3
603.9
3,333
3,170
3,150
152.4
145.0
144.0
Year
NOTE: CPI deflator is BLS’ water and sewer deflator for all urban consumers.
SOURCE: Revenue taken from Census of Governments, annual survey
In short, there are many problems and questions surrounding the use of the CPI as a
deflator to compute a water output index. The primary advantage of the CPI deflator is
that it is available monthly, quarterly, and annually. It reflects, at least in a general way,
the change in water prices over the past 25 years. Mindful of its strengths and weaknesses, this bulletin uses it to calculate a water output index.
Water output index calculations show an increase in deflated revenue of 44 percent
over the 1967-92 period. The average annual increase is 1.5 percent. However, the
periods of growth have not been uniform. There was a sizable spurt during the 1980s,
but since 1989 there has been a decrease (table 22).
Output index adjusted for quality. The Consumer Price Index shows that the unit price
of drinking water rose about 500 percent between 1967 and 1992 (table 22). Most of
the increase, apparently, is due to inflation brought about by increases in the cost of
factor inputs, but part is due to other considerations. They include: Quality improvements resulting from increased treatment; attempts to capture the entire cost of production whereas in the past many water utilities have been subsidized by general taxes; and
attempts to reduce water consumption by raising the unit price as water consumption
increased. Ideally, each factor should be isolated and an assessment made how each
affects the growth of output. But by using the CPI deflator, all these factors are removed when in fact it is desirable to remove only the price increase.
Enterprise Service—56
In this section an attempt is made to identify that part of the increase due to quality
improvements and adjust the output index accordingly. The procedure used is to identify the funds that the utilities spent on water testing and treatment, deflate them, and
add them to the deflated output series. In an imperfect sense the resources are captured
that have been devoted to improving the quality of the water.
This procedure is dictated largely by data availability. Two primary sources of data
are used, the EPA estimates of funds spent on testing and treatment of water, and the
BEA deflators for water and sewer pollution abatement. For 1967-72, no adjustments
were made because EPA made no estimates. EPA claims that its research suggests that
expenditures for treatment increased at about the same rate as did overall expenditures
for water utilities during this period. For 1972-92, treatment expenditures were taken
from EPA; for 1972-1987 the statistics are estimates, for 1988-92 they are projections.59
The real increase in testing and treatment came after the passage of the Safe Drinking
Water Act of 1976. The deflators were taken from recent estimates by BEA for 197292.60
The results of these calculations show that the quality adjusted index increased 54
percent while the unadjusted index increased 44 percent. The average annual increase
for the index adjusted for quality is 1.7 percent; for the unadjusted index it is 1.5 percent (table 23).
Adjustments for quality and product differentiation. The index was also adjusted to
reflect the shifts among the different services or customers. For example, there has been
a relative shift from industrial to residential sales as industrial users have reduced their
water consumption. To capture this shift, the overall index was divided into the five
customer groups previously noted using sample data. Each group was weighted with
the appropriate average revenue weights also using sample data and the segments were
linked. The base years are 1967, 1970, 1976, and 1984, which reflect data availability.
The results of these calculations show an overall increase of 57 percent versus 54
percent for undifferentiated sales over the 1967-92 period. The average annual increase was 1.8 percent versus 1.7 percent for the quality adjusted index and 1.5 percent
for the unadjusted index (table 23).
Comparison with other indexes. Because of the numerous questions surrounding the
deflated value index, it was compared with six other indexes. They are: A sample of
revenue produced water (or gallons sold) for 1965-81 drawn from American Water
Works Association (AWWA) statistics; a sample of the gallons of water pumped for
1965-84 (AWWA) statistics; a sample of revenue (deflated with the CPI water and
sewer deflator) for 1965-84 (AWWA) statistics; a sample of the number of connections
for 1965-84 (AWWA); public gallons supplied as collected by the U.S. Geological
Survey for 1965, 1970, 1975, 1980, 1985, and 1990; and domestic gallons produced as
collected by the U.S. Geological Survey for the same 5 years. These six indexes are all
undifferentiated and none are adjusted for changes in water quality. Thus, the comparison is with the unadjusted, undifferentiated deflated value index presented earlier (table
23).
The four American Water Works Association samples were constructed from data
collected by the Association in 1965, 1970, 1976, 1978, 1981 and 1984. The statistics
collected vary by survey, but the data were published by utility. The procedure used to
construct the comparison indexes was to include all utilities that employed 250 or more
FTE employees in any sample year. This resulted in a sample of 49 utilities including
those in most large cities. While the sample is probably not representative of U.S.
water utilities, it does include all the large utilities and much of the U.S. population.
Two indexes were constructed from data taken from U.S. Geological Surveys that it
conducts every 5 years. One index, public supply, reflects water supplied by public and
private water groups. It includes domestic, commercial, industrial, thermoelectric power,
Enterprise Service—57
Table 23. Three local government water utility output indexes, 1967-92
(1967=100)
Quantity
Quantity/
quality
Quantity/quality
productiondifferentiation
1967 ......................................
1968 ......................................
1969 ......................................
100.0
101.1
101.0
100.0
101.1
101.0
100.0
101.6
101.9
1970 ......................................
1971 ......................................
1972 ......................................
1973 ......................................
1974 ......................................
1975 ......................................
1976 ......................................
1977 ......................................
1978 ......................................
1979 ......................................
102.1
102.2
104.9
108.5
109.7
111.5
108.3
109.3
108.4
117.4
102.1
102.2
104.9
108.4
109.5
111.8
109.2
111.1
110.2
120.3
103.5
103.7
106.6
110.3
111.6
114.1
111.5
113.5
111.8
122.1
1980 ......................................
1981 ......................................
1982 ......................................
1983 ......................................
1984 ......................................
1985 ......................................
1986 ......................................
1987 ......................................
1988 ......................................
1989 ......................................
119.0
121.2
118.8
123.7
127.5
137.3
144.1
148.5
149.7
154.4
121.9
124.5
123.3
128.2
131.9
141.8
148.9
153.9
154.7
159.5
123.8
126.5
125.4
130.5
134.4
144.4
151.7
156.7
157.6
162.5
1990 ......................................
1991 ......................................
1992 ......................................
152.4
145.0
144.0
158.1
151.6
154.1
161.0
154.5
156.9
Average annual
rate of change:
1967-92: ...............................
1.5
1.7
1.8
Year
NOTE: The first index is deflated value only; the second is deflated value adjusted for quality,
but not product differentiation; the third index is deflated value adjusted for changes in quality and
product differentiation.
and public water. The other index, domestic supply, is water for household purposes; it
is also called residential water and may be drawn from a public supply or may be selfsupplied—e.g., private pump. Both series include private as well as government supplies.61
Comparison of these six indexes and the deflated value index presented above show
that they all increased during the measured period. Because the six comparison indexes
and the deflated value indexes did not cover the same time period it was necessary to
examine four separate time periods, 1965-70, 1970-76, 1976-81, and 1981-84. These
comparisons showed reasonably good matches in all periods except 1965-70.
The USGS indexes are the only two comparison indexes for which data are available
for most of the period (table 24). They include data for 1965, 1970, 1975, 1980, 1985,
and 1990. Indexes computed from these data show that from 1965-90, the USGS domestic index increased at an average annual rate of 2.0 percent, and the USGS public
index increased at a 1.9 percent-rate. The deflated Census revenue index increased at
1.8 percent. However, there was considerable divergence between the USGS indexes
and the deflated value index during the individual segments, particularly in the early
years.
In summary, although there are questions concerning the Census of Governments’
deflated value series, as there are with all the data, the deflated value index generally
moves in concert with the other indexes over the long term.
Enterprise Service—58
Table 24. Comparison of three water supply output indexes, 1965-90
(1965=100)
Year
Public
supply
Domestic
supply
Deflated
value
1965 ..............................
1966 ..............................
1967 ..............................
1968 ..............................
1969 ..............................
100.0
100.0
100.0
102.1
103.0
104.2
104.0
1970 ..............................
1971 ..............................
1972 ..............................
1973 ..............................
1974 ..............................
1975 ..............................
1976 ..............................
1977 ..............................
1978 ..............................
1979 ..............................
114.2
119.0
122.4
124.4
105.2
105.3
108.0
111.7
113.0
114.9
111.6
112.6
111.6
120.9
1980 ..............................
1981 ..............................
1982 ..............................
1983 ..............................
1984 ..............................
1985 ..............................
1986 ..............................
1987 ..............................
1988 ..............................
1989 ..............................
143.4
137.0
153.9
152.0
1990 ..............................
162.4
158.4
157.0
Average annual
rate of change:
1965-90 .........................
2.0
1.9
1.8
122.5
124.8
122.4
127.4
131.3
141.4
148.4
152.9
154.2
159.0
SOURCE: Public and domestic supply taken from USGS reports; Puerto Rico and
Virgin Islands removed from the totals. Deflated value computed; no adjustments for
quality or product
Labor inputs
There were almost 156,000 local government water utility employees in 1992, a 35percent increase over 1967. Municipalities employ the majority (69 percent) of these
workers. The rest work for special districts (21 percent), counties (7 percent), and
townships (3 percent).62
The average number of full-time-equivalent employees per utility is about 10, however, the numbers range from zero to thousands. Los Angeles had the most with almost
3,400, followed by New York City with 2,900 and Chicago with 2,100.63 Some of the
small utilities use contract or volunteer personnel.
Most (92 percent) water employees are full-time workers. Comparable figures for
gas and electric are 96 and 97 percent, respectively. Like gas and electric, the percent
of full-time water employees has remained relatively constant over time. In 1967, for
example, 93 percent worked full time, in 1977 it was 93 percent, in 1987 it was 94
percent, and in 1992 it was 92 percent.
The number of full-time-equivalent local government employees was almost 147,000
in 1992. This number has increased at about the same rate as total employment. The
FTE increase between 1967-92 is 36 percent while the increase in total employment is
34 percent. The average annual increase over this period is the same for both indexes,
1.2 percent.
Enterprise Service—59
As discussed in chapter 2, two labor indexes, FTE and total employment, are usually
calculated for government productivity measures. Occasionally the two indexes move
in different directions as employees work increasing amounts of overtime or part-time
employees are substituted for full-time staff. But, in most cases they move in lock-step,
as is the case with drinking water (table 25).
Table 25. Total and FTE employment, local government
water supply indexes, 1967-92
(1967=100)
Year
Total
FTE
employees
1967 ...........................
1968 ...........................
1969 ...........................
100.0
99.5
101.2
100.0
99.3
101.8
1970 ...........................
1971 ...........................
1972 ...........................
1973 ...........................
1974 ...........................
1975 ...........................
1976 ...........................
1977 ...........................
1978 ...........................
1979 ...........................
103.2
103.1
101.9
106.5
113.4
112.4
110.8
111.8
114.5
113.8
104.0
103.8
104.4
107.8
112.4
111.8
110.6
112.6
115.8
113.3
1980 ...........................
1981 ...........................
1982 ...........................
1983 ...........................
1984 ...........................
1985 ...........................
1986 ...........................
1987 ...........................
1988 ...........................
1989 ...........................
115.0
115.5
117.1
119.7
116.4
120.1
121.0
123.6
125.6
127.5
115.7
116.4
119.6
121.8
119.5
121.3
124.1
126.1
127.5
129.0
1990 ...........................
1991 ...........................
1992 ...........................
128.7
131.8
134.4
130.6
133.4
136.0
Average annual
rate of change:
1967-92 ......................
1.2
1.2
The indexes presented in table 25 were constructed from the annual Bureau of the
Census sample of local governments, and were benchmarked to data collected by that
agency in its 100 percent sample taken every 5 years. Although the trends computed
from the sample and the benchmarked data are similar, the employment levels are consistently higher for the sample. Overall sample results are 3.9 to 8.4 percent higher than
the benchmark numbers. It is not known why there is such a consistent bias. Other
local government services, such as gas, electric, and alcoholic beverage sales, reveal
sample variance from benchmark year to benchmark year, but not the consistent bias
found here.
Both the total employment and FTE employment indexes are presented in table 25.
Because the two indexes parallel each other, only the FTE is discussed in this section.
Over the period 1967-92, the average annual increase in the FTE index was 1.2
percent; increases occurred in 19 of the 25 years. There has been no drop in employment since 1984.
Enterprise Service—60
Productivity indexes
The average annual increase of 1.2 percent for labor and 1.8 percent for output over
the 1967-92 period results in an average annual increase in labor productivity of 0.6
percent. However, the overall increase in the productivity index is marked by numerous, minor fluctuations; productivity during the last three years (1989-92) decreased at
an average annual rate of 2.9 percent (table 26).
These statistics reflect the differentiated output index adjusted for quality changes
and the FTE employment indexes. If the output index is not adjusted for change in
quality and product mix there is still a small (0.2 percent) increase. But, there are some
fairly large fluctuations, and for almost half the measured years there was a decrease in
labor productivity.
Despite the questions surrounding the measurement of water utility outputs, it does
appear that there has been a small increase in local government water utility labor productivity over the 1967-92 period.
Table 26. Local government water supply productivity indexes,
1967-92
(1967=100)
Year
Output
Input
Productivity
1967 ........................
1968 ........................
1969 ........................
100.0
101.6
101.9
100.0
99.3
101.8
100.0
102.3
100.1
1970 ........................
1971 ........................
1972 ........................
1973 ........................
1974 ........................
1975 ........................
1976 ........................
1977 ........................
1978 ........................
1979 ........................
103.5
103.7
106.6
110.3
111.6
114.1
111.5
113.5
111.8
122.1
104.0
103.8
104.4
107.8
112.4
111.8
110.6
112.6
115.8
113.3
99.5
99.9
102.1
102.3
99.3
102.0
100.8
100.8
96.6
107.8
1980 ........................
1981 ........................
1982 ........................
1983 ........................
1984 ........................
1985 ........................
1986 ........................
1987 ........................
1988 ........................
1989 ........................
123.8
126.5
125.4
130.5
134.4
144.4
151.7
156.7
157.6
162.5
115.7
116.4
119.6
121.8
119.5
121.3
124.1
126.1
127.5
129.0
107.0
108.6
104.8
107.2
112.4
119.0
122.2
124.2
123.6
125.9
1990 ........................
1991 ........................
1992 ........................
161.0
154.5
156.9
130.6
133.4
136.0
123.3
115.8
115.4
Average annual
rate of change:
1967-92 ...................
1967-89 ...................
1.8
2.2
1.2
1.2
0.6
1.1
SOURCE: Tables 23 and 25
Enterprise Service—61
Mass Transit
Mass transit is categorized as a utility by the Bureau of Census and as an enterprise
service by the Bureau of Economic Analysis. Nevertheless, it differs from electric
power, natural gas, and water supply in several important respects:
•
•
•
It is labor intensive;
Its fees and charges fail to support the service;
It is heavily subsidized by Federal, State, and local government.
Institutional setting
State and local government delivered 94 percent of the transit industry’s passenger
trips, operated 94 percent of its vehicle miles, and owned or leased 86 percent of its
vehicles in 1990.64 Today, mass transit is government transit. However, this has not
always been the case.
A dramatic shift has occurred in transit ownership and operation over the past 50
years: State and local government has stepped in to operate failed private systems. By
one count, only 36 systems, or about 3 percent of all transit systems in the United States
were government owned and operated in 1950. By 1980, 576 systems were government owned and today most medium and large-size systems are public. In 1950, 28
percent of the industry’s vehicles were publicly owned or leased; in 1970, the figure
was 66 percent and in 1990 it was 86 percent. In 1970, 68 percent of all vehicle miles
were public; in 1980 the figure was 93 percent and in 1990, as noted above, it was 94
percent.65
Most transit service is government supplied either through direct provision or by
contract with private firms. There are a variety of contractual agreements ranging from
private operation with government subsidies to total government operation. The indexes presented in this section reflect the service provided by government employees.
State and/or local government transit systems operated in every State in 1992. New
York, with about 65,000 State and local government employees and $10.1 billion in
expenditures, is the most deeply involved. The large systems are concentrated in urban
areas, particularly in the Northeast, and they dominate production. The 5 largest systems accounted for 38 percent of all transit employment in the Nation; the largest 10
account for 46 percent; and the largest 13 account for 50 percent.
In fiscal 1992, State and local transit systems spent approximately $21.9 billion, or
almost 2 percent of all State and local government expenditures. However, their capital
expenditures accounted for more than 4 percent of all State and local capital expenditures.66
Passenger fares covered about 36 percent of all transit operating expenditures in
1992. Transfer payments and gasoline, sales, and property taxes support the other 64
percent. The Federal Government contributed 5 percent of all operating expenditures,
and almost 50 percent of all capital expenditures.67
The primary factor input into transit operations is labor. According to one set of
data, salaries, wages and benefits accounted for about 42 percent of all transit expenditures (capital and current operations), and 61 percent of all current operating expenditures in 1992 (table 27). A different data set, which includes private as well as public
systems and captures the wages, salaries and fringe benefits paid to contractor employees, shows that compensation accounts for about 73 percent of all current operating
costs.68
Public transit consists of a variety of operational modes. There are six basic types of
public transport: Bus, heavy rail such as subway, commuter rail, light rail such as streetcar, trolley bus, and demand response. Less common types include: Urban ferry boat,
cable car, incline plane, aerial transport, and several other modes. Bus and heavy rail
dominate the mass transit industry carrying about 90 percent of all passengers and producing 82 percent of all vehicle revenue miles (table 28).
Enterprise Service—62
Table 27. Finances of State and local government mass transit by type of
government, fiscal year 1992
(millions of dollars)
Category
Total
Revenue .................. $5,742
Expenditures ........... 21,879
Capital ................ 5,836
Interest on debt ..
968
Current
operations ...... 15,076
Compensation .. 9,150
Other ................ 5,926
States
Counties Municipali- Townships Special
ties
Districts
$1,126
4,292
1,713
257
$165
830
200
11
$1,892
5,613
990
230
$1
5
—
—
$2,559
11,139
2,933
470
2,322
1,131
1,191
620
370
250
4,392
2,886
1,506
5
3
2
7,736
4,760
2,976
NOTE: “Compensation” estimated by multiplying Bureau of Census October pay times 12.
Standard benefits were derived for general government and applied to salary and wage estimates
to reach total compensation. “Other” is a residual—current operations minus compensation. Because of rounding, detail may not add to total.
SOURCE: U.S. Bureau of the Census, Government Finances: 1991-92, Series GF/92-5. U.S.
Government Printing Office, Washington, DC (1996) and U.S. Bureau of the Census, Public Employment: 1992, Series GE/92-1. U.S. Government Printing Office, Washington, DC (1994)
Table 28. Transit modes (private and public) ranked by passenger
trips, vehicle revenue miles, and operating expense, 1992
(in percent)
Mode
Passenger
trips
Vehicle
revenue
miles
Operating
expenses
Total ..................................
Motor bus ..............................
Heavy rail (subway) ...............
Commuter rail .......................
Light rail (streetcar) ...............
Trolley bus .............................
Demand response ................
Other .....................................
100.0
61.7
28.7
4.1
2.4
1.5
.6
1.0
100.0
61.4
20.1
7.9
1.1
.4
8.2
.9
100.0
55.7
22.9
14.0
2.0
.7
3.2
1.5
SOURCE: Computed from U.S Federal Transit Administration, National Transit
Summaries and Trends, (from the 1992 database), p. 15; And 1993 Transit Fact
Book, (American Public Transit Association), pp 27-28.
The motor bus, the most important mode in 1992, accounted for 62 percent of all
transit passenger trips, 61 percent of the vehicle revenue miles, and 56 percent of all
operating expenses (table 28). Over 50,000 urban motor buses operated in the United
States in 1992. These vehicles logged over 4.7 billion passenger trips and approximately 1.6 billion vehicle revenue miles. Over half of all passenger trips were logged
by the 15 largest bus systems, and about 40 percent of the bus revenue hours and miles.69
Heavy rail or subway is the next most important mode of public transit. In 1992,
heavy rail accounted for 29 percent of all passenger trips, 20 percent of vehicle revenue
miles, and 23 percent of operating expenditures. The 13 heavy rail systems, all publicly
owned and operated, logged 2.2 billion passenger trips for an estimated 10.7 billion
passenger miles. New York City dominates the heavy rail systems, accounting for over
half of all U.S. passenger trips, passenger miles, vehicle revenue miles, and vehicle
revenue hours.70
The 18 commuter railroads logged about 314 million passenger trips and 7.3 billion
passenger miles in 1992. All were government owned or received government subsidies, but most are operated under contract. Amtrak, for example, provided service for
Los Angeles and San Diego, and the Baltimore and Ohio provided service to Baltimore
and Washington.71
Enterprise Service—63
Sixteen light rail systems, of which 15 were government, were operated in 1992.
This mode shows substantial growth since the early 1970s. In 1992, the 16 light rail
operations carried 187 million passengers a total of 700 million passenger miles. The
six largest systems accounted for about 75 percent of the passenger trips and passenger
miles.72
Only five trolley systems operated in 1992, the same number as in 1980. A relatively minor mode in terms of mass transit, it accounted for only 126 million passenger
trips and 199 million passenger miles in 1992. The San Francisco system accounts for
over half the country’s trolley service.73
Demand response accounted for 45 million passenger trips and 209 vehicle revenue
miles in 1992, much of it operated by small, private firms under contract to government
agencies. Formerly known as dial-a-ride, it is a service that has grown dramatically
over the past decade. In earlier years it was not even identified as a separate mode, but
simply lumped under the general heading of paratransit along with jitneys and airport
limousines.74
The following discussion focuses on bus and heavy rail operations because of their
importance. The tabulations, however, include light rail and trolley operations, but
exclude demand response, commuter rail, and the lessor modes of transportation. School
buses, which are normally operated by school systems, also are excluded.
This discussion focuses on fixed route systems as defined in the Standard Industrial
Classification Manual (SIC 4111):
Establishments primarily engaged in furnishing local and suburban mass
passenger transportation over regular routes and on regular schedules, with
operations confined principally to a municipality, contiguous municipalities, or a municipality and its suburban areas. Also included in this industry
are establishments primarily engaged in furnishing passenger transportation by automobile, bus, or rail to, from or between airports or rail terminals, over regular routes, and those providing bus and rail commuter services.75
Extensive data collection and research support transit, probably more than any other
government service. The American Public Transit Association (APTA) and the Bureau
of Census have collected data on individual transit systems and produced national statistics for many years. The U.S. Department of Transportation also collects very detailed statistics on individual transit systems using comparable definitions and reporting forms. These are known as the National Transit Database. Using them, and other
data, the research community has conducted extensive research on the operations of the
service including studies of productivity. In addition, there have been demonstration
projects and evaluations.
Outputs
Over the past several decades transit output measures have been expanded and massive amounts of data have been collected, edited, and published. There is no lack of
measures and measurements of public transit operations today.76 The output measures
that seem to be most often recommended for productivity measurement purposes include: Vehicle revenue hours, vehicle revenue miles, vehicle miles, number of passengers, passenger miles, and passenger revenue.
Private sector analysts often focus on number of passengers, passenger miles, and
passenger revenues, while public sector transit analysts and managers use all the measures, but focus on vehicle revenue hours, vehicle revenue miles, and vehicle miles for
productivity calculations. The strengths and weaknesses of each of these measures are
analyzed at length in the literature. The following discussion briefly reviews this research.
Vehicle revenue hours. Vehicle revenue hours (VRH) seems to be the output measure
that most transit managers prefer for productivity analysis. Revenue refers to the hours
Enterprise Service—64
a vehicle is in service and capable of generating revenue, not the amount of revenue
actually generated. VRH do not include the hours spent traveling to and from storage
facilities, other deadhead travel, or layover time. Vehicle revenue hours is a measure of
transit availability or capacity. A bus could travel a route for 8 hours without any
passengers but still generate 8 revenue hours.
Arguments in favor of this measure are that it is measurable, repetitive, accurate,
easily understood, a good measure of the cost of production, and data are readily available with which to calculate this index. It is also a measure over which decision makers
have good control, it encourages reduction of nonproductive use of vehicle deadhead
and layover time, and it is a good measure of the service provided to a community. If
the hours of service are extended, revenue vehicle hours increase. If service is cut, they
decrease.
The principal argument against using vehicle revenue hours as an output measure is
that it is a measure of capacity, not use. An increase in vehicle revenue hours does not
necessarily lead to an increase in the number of passengers carried or revenue collected. It is not a measure of final output, at least in the context in which the term is used
here.
Vehicle revenue miles. This is a measure of the miles traveled by vehicles while picking
up and transporting passengers. It does not include miles driven to start a route such as
deadhead miles. It is analogous to VRH except that it uses miles rather than hours.
There should be a very high correlation between the two measures. The arguments for
and against using vehicle revenue miles as an output measure are similar to those already discussed under vehicle revenue hours. The primary argument in favor of miles
is its availability.
Vehicle miles. This is a measure of the number of miles traveled by transit vehicles. It
should correlate highly with vehicle revenue miles and vehicle revenue hours. Both of
these measures are more useful for productivity measurement purposes because they
exclude non-productive time. Nevertheless, the argument in favor of using vehicle
miles instead of revenue vehicle miles is data availability. Most, if not all, transit systems have collected data on vehicle miles for years.
Passenger trips. A measure of transit use, in contrast to capacity provided, is the number of passengers or passenger trips. The basic measure is a count of all passengers
using a transit system. The measure is sometimes divided by type of passenger—paying, nonpaying, school child, reduced fare, elderly, and so forth. A special concern is
how to count passenger transfers. Some systems count transfers as additional passengers; others do not.
The variation of this measure that is used most frequently today is the unlinked
passenger trip (UPT) which is defined as the number of transit vehicle boardings. Each
time a person boards a vehicle he or she is counted; each transfer as well as the initial
boarding is counted as an unlinked passenger trip.
The basic strength of the unlinked passenger trip measure is its focus on usage. “For
the typical transit system, increased patronage from one year to the next is much more
significant than any other financial or operating statistic.”77 Most, if not all, transit
systems keep statistics on the number of passengers. Also, the measure is a physical
count, is easily understood, is measurable, and it is repetitive.
The public transit industry’s basic problem with this measure for productivity assessment is that it is an effectiveness or outcome indicator of the service provided, not
a measure of final output. This argument applies equally to all consumption-based
measures of transit output. Other arguments against this measure are that some transit
systems do not accurately measure the number of passengers, and that this measure
provides no information about the length of the ride.
Passenger miles. Passenger miles are probably the most widely used physical output
measure of private sector passenger transportation productivity. This measure is supeEnterprise Service—65
rior to a count of the number of passengers because it takes into account the length of
the ride which is important if the length of the ride is changing. Passenger miles are
normally defined as the number of miles traveled by all paying passengers in a set time
period. One passenger traveling one mile is one passenger mile. The private sector
studies of Deakin and Seward; Kendrick; Scheppach and Woehlcke; and the U.S. Bureau of Labor Statistics all use passenger miles in one form or another.78
The arguments normally advanced against passenger miles as the measure of output
generally are the same as those noted in the case of the number of passengers. Probably
the most fundamental criticism is data availability and accuracy. A primary problem is
estimating how far passengers ride in a fixed-fare system. If the length of the average
passenger trip remains constant through time, which is probably a reasonable assumption in the short run, then passenger miles and the number of passengers would result in
the same index.
Passenger revenue. This measure is the total revenue collected from passengers. Passenger revenue, sometimes known as “farebox revenue,” is available in every system
that collects fares, and is available nationally. For those systems that cover costs through
fares, it reflects transit usage. Also, it is repetitive and accurate, and it is easily understood.
For several reasons, passenger revenue is not a particularly good measure of output
for productivity measurement. First, to calculate an output index would require an
accurate set of deflators. Second, passenger revenue comprises less than 40 percent of
national transit revenue (subsidies make up the rest), and it has decreased as a percent
of total revenue over the years. Hence, even with a good set of deflators, a deflated
revenue index would not be a good measure of transit output. Third, nonpaying passengers are often an important user group. Every system has some, and some systems have
many. A few systems charge no fares whatsoever. Finally, passenger revenue is a
function of administered fares, which may or may not be a function of the cost of providing the service.
Which measure?
If one views transit as an enterprise service, then usage should be the output measure. That is, if it is categorized as a private good like electric power and drinking water,
and it supports its operations by fees and charges, then the focus should be on the
service consumed.
If, on the other hand, transit is categorized as a general government service, and
there are arguments that suggest that this should be the case, then a capacity measure
might be preferred. It is clear that the public transit industry prefers this approach.
Rather than wrestle further with the capacity-usage conundrum, this bulletin computes one measure of each. They are complementary measures and provide very different insights into transit output and productivity, and for this reason alone they should
both be calculated.
Unlinked passenger trips is the usage measure chosen for examination. Although the
concept of passenger miles is preferred conceptually, good data are lacking, particularly in the early years. Even in later years, when passenger mile data are available,
there are questions concerning their accuracy. As noted above, passenger mile and
passenger trip indexes should track reasonably well in the mass transit environment, at
least over the short run.
The chosen capacity measure is vehicle revenue miles. Vehicle revenue hours is the
preferred conceptual measure, but data to calculate the measure are more readily available for VRM, particularly in the early years.
Data to calculate each output measure should be categorized by service mode, such
as bus or heavy rail, because of different factor inputs. Bus and heavy rail dominate
transit service. The indexes prepared as part of this bulletin include four modes—bus,
heavy rail, light rail, and trolley. These accounted for more than 80 percent of all transit
employment in 1992.79
Enterprise Service—66
Quality and level of service
The quality and level of service are important considerations in every transit operation. Travel time, reliability, safety, and comfort are all important dimensions of transit
output. Chapter 2 suggested that such factors need to be explicitly considered and
adjustments made whenever they affected base-year unit labor requirements or cost
weights. Such requirements will be affected by three considerations: The relationship
of quality and level of service to total labor or cost requirements; the unit output measure; and the magnitude of the change.
Quality and level of service should be taken into account when a change markedly
affects labor or cost requirements. This is not to say that the attribute is unimportant
when it does not affect these factors, only that it need not be considered in productivity
calculations. For example, employee courtesy is an important quality attribute, to which
the public and transit authorities, alike are sensitive. To improve employee courtesy,
transit managers sponsor “Driver of the Month” awards and courtesy training, and when
all else fails they may use disciplinary action. Important as these programs might be to
a transit manager, they are relatively unimportant for resource requirements, and need
not be considered in transit productivity calculations.
The second factor to be considered is the output measure used to calculate productivity. A change in a quality or level of service attribute can affect unit costs for some
measures but not others. For example, frequency of service plays a major role in unit
costs. Systems that provide 24-hour service usually have very different unit cost requirements from those that provide only rush hour service. Changing the level of service will likely have a major effect on unit costs for output measures such as passenger
miles, but will have little effect on vehicle revenue hour measures.
The third point is that, for trend calculation, adjustments are not needed as long as
quality and level of service remain constant or approximately constant. To make such
a judgment requires that these attributes be followed through time.
Productivity literature often cites seven quality and level of service attributes: Courtesy, comfort, safety including accidents, security including crime, convenience and
accessibility, reliability, and travel time. By calculating possible program changes in
each of these areas their impact on unit labor requirements can be estimated. This will
also provide information which might usefully be examined and tracked through time.80
This review suggests that within a reasonable domain, changes in four attributes—courtesy, safety, reliability, and travel time—are not likely to affect unit labor requirements.
However, for three—comfort, convenience and accessibility, and security unit labor
requirements could be affected.
The quality attributes most susceptible to programmatic changes are:
•
•
•
•
Air conditioning of vehicles and stations (comfort);
The Americans with Disabilities Act (ADA) requirements which include
wheelchair lifts on vehicles and station elevators (convenience and accessibility);
Hours of service (convenience and accessibility);
Change in the number of transit police (security).
National time series data are generally lacking with which to assess changes in these
attributes. But the available statistics suggest that there has been little change in the
ratio of transit police to operating personnel from 1967-92, and the hours of operation
have not been modified, at least in the large systems. The ADA requirements are likely
to have an affect, but through 1992 it has been mostly in capital improvements. Air
conditioning has been installed in most transit system vehicles over the past 25 years,
which certainly has affected unit cost and labor requirements. However, the degree of
impact is not known, and no modification of the data have been attempted to take this
into account.
Enterprise Service—67
Index construction and data
availability
Productivity calculations in this chapter focus on two output measures, the unlinked
passenger trip index and the vehicle revenue mile index. For each, separate indexes of
bus, heavy rail, light rail, and trolley operations have been calculated. The individual
modes were then combined using base period unit labor weights with 1967, 1972, 1977,
1982 and 1987 as the reference years.
The data to calculate the output indexes were taken primarily from the U.S. Department of Transportation National Transit Database for 1979-92, and APTA Transit Operating Reports, individual system annual reports, and special studies for the 1967-79
period. The statistics reflect all heavy rail, light rail, and trolley facilities operated by
local government employees in the base year. Bus operations include all local transit
systems with 200 or more full-time-equivalent employees. In 1967, only 21 systems
were included; in 1987, 70 were included. Not included in these statistics are cable car,
commuter rail, demand response, ferry boat, incline plane, and monorail facilities. The
1992 sample covered 83 percent of all local government mass transit employees.81
Vehicle revenue mile
indexes
The number of local government mass transit vehicle revenue miles increased by
more than 75 percent between 1967 and 1990; it has since dropped slightly. The average annual increase in VRM between 1967 and 1992 is 2.2 percent, with VRM increasing in 20 of the 25 measured years. The largest increase came early in the period when
there was a rapid shift from private to government operations. Between 1967 and 1977
the average annual increase was 4.2 percent; between 1982 and 1992 it was less than 1
percent per year (table 29).
In the modal series, vehicle revenue miles increased for bus, heavy rail, and light rail
while trolley dropped by almost half. The largest percentage increase was in light rail
as new systems were built and introduced into operation. The average annual increase
in light rail VRM was 5.7 percent. Bus increased at an average annual rate of 2.8
percent and heavy rail increased at 1.2 percent. But from 1982-92 there was virtually
no change in bus VRM in the sample, while heavy rail and light rail VRM increased
rapidly.
Enterprise Service—68
Table 29. Vehicle revenue mile indexes for local government mass transit
service by mode, 1967-92
(1967=100)
Year
Total
Bus
Heavy
rail
Light
rail
Trolley
1967 .......................
1968 .......................
1969 .......................
100.0
101.7
104.2
100.0
101.3
103.0
100.0
102.7
107.2
100.0
98.9
97.7
100.0
96.9
87.8
1970 .......................
1971 .......................
1972 .......................
1973 .......................
1974 .......................
1975 .......................
1976 .......................
1977 .......................
1978 .......................
1979 .......................
118.0
120.8
122.6
132.7
137.2
144.5
150.9
150.5
150.1
151.3
120.9
124.3
129.6
151.9
159.2
175.0
188.7
192.6
192.7
193.3
114.7
116.5
113.7
108.8
110.1
106.2
103.1
96.8
95.4
97.6
211.6
238.5
238.5
238.5
239.7
231.3
231.0
227.3
242.9
244.0
87.5
84.3
82.3
65.2
59.9
60.4
58.5
56.2
55.7
55.4
1980 .......................
1981 .......................
1982 .......................
1983 .......................
1984 .......................
1985 .......................
1986 .......................
1987 .......................
1988 .......................
1989 .......................
155.6
160.1
160.2
157.3
160.7
162.0
165.5
166.0
170.4
176.6
197.4
202.9
203.2
197.4
198.4
197.5
200.0
198.6
201.0
209.3
102.5
105.9
105.6
106.4
113.5
117.8
123.1
126.2
133.2
137.0
241.2
259.7
250.6
241.2
246.2
236.7
249.4
269.5
304.6
311.9
56.1
52.2
55.4
59.0
60.3
61.5
58.1
58.9
57.7
56.9
1990 .......................
1991 .......................
1992 .......................
177.0
174.0
172.1
208.7
204.8
201.7
138.2
135.7
135.2
354.4
404.2
401.0
54.1
53.4
54.8
Average annual
rate of change:
1967-92 ..................
2.2
2.8
1.2
5.7
-2.4
Enterprise Service—69
Passenger trip indexes
The increase in unlinked passenger trips (UPT) is much more modest. The average
annual increase between 1967 and 1992 for all four modes combined is 0.5 percent.
While there has been a steady increase in vehicle revenue miles, the UPT index has
generally declined since 1984 (table 30).
The decrease in unlinked passenger trips is reflected in modal statistics. Both heavy
rail and trolley show decreases in UPT between 1967 and 1992, and bus travel dropped
between 1981 and 1992. Only light rail passenger miles have shown an overall increase
through the entire measured period.
Table 30. Unlinked passenger trips of local government mass transit by mode,
1967-92
(1967=100)
Year
Enterprise Service—70
Total
Bus
Heavy
rail
Light
rail
Trolley
1967 .......................
1968 .......................
1969 .......................
100.0
100.4
101.1
100.0
100.6
101.1
100.0
100.5
102.1
100.0
98.3
98.3
100.0
93.0
85.2
1970 .......................
1971 .......................
1972 .......................
1973 .......................
1974 .......................
1975 .......................
1976 .......................
1977 .......................
1978 .......................
1979 .......................
107.5
104.3
102.1
104.2
106.2
110.3
111.1
110.7
112.2
121.0
108.8
107.4
106.3
113.7
118.6
127.0
130.6
130.2
130.4
143.6
105.0
99.7
96.0
92.1
90.8
90.0
87.1
86.6
89.4
93.1
194.9
193.0
188.4
188.3
188.1
178.8
175.2
175.6
191.4
193.4
88.3
79.9
77.7
67.3
62.4
60.4
60.3
62.9
64.8
66.7
1980 .......................
1981 .......................
1982 .......................
1983 .......................
1984 .......................
1985 .......................
1986 .......................
1987 .......................
1988 .......................
1989 .......................
125.8
126.4
121.8
119.2
125.1
123.1
121.4
119.7
118.3
120.1
150.6
152.1
144.2
136.5
145.2
140.0
135.6
130.5
130.9
132.6
93.6
93.0
91.0
94.3
97.1
100.1
102.1
104.4
100.3
102.2
211.7
213.2
218.1
231.1
224.8
221.7
209.7
221.4
241.8
255.9
86.7
87.6
106.8
109.4
112.7
99.0
97.3
98.5
95.0
90.8
1990 .......................
1991 .......................
1992 .......................
116.1
111.3
113.4
131.1
127.5
124.6
94.3
87.3
95.9
276.5
290.7
299.9
88.0
87.3
87.9
Average annual
rate of change:
1967-92 ..................
0.5
0.9
-0.2
4.5
-0.5
Weighted output
indexes
To capture modal shifts in transit service through time, weighted output indexes
were constructed for both the VRM and UPT series. The process was discussed earlier
in this section and in chapter 2. Examination of the weighted and unweighted indexes
shows remarkably little difference, a reflection of the dominance of bus and heavy rail
in the transit series, and the absence of major shifts in the unit labor requirements through
time (table 31).
Table 31. Weighted and unweighted output indexes for vehicle revenue
miles and unlinked passenger trips in local government mass transit
service, 1967-92
(1967=100)
Vehicle revenue miles
Unlinked passenger trips
Weighted
Unweighted
Weighted
Unweighted
1967 .......................
1968 .......................
1969 .......................
100.0
101.7
104.2
100.0
101.7
104.2
100.0
100.3
101.1
100.0
100.4
101.1
1970 .......................
1971 .......................
1972 .......................
1973 .......................
1974 .......................
1975 .......................
1976 .......................
1977 .......................
1978 .......................
1979 .......................
118.6
121.5
123.2
131.8
136.0
142.3
147.8
146.7
146.4
147.7
118.0
120.8
122.6
132.7
137.2
144.5
150.9
150.5
150.1
151.3
107.8
104.6
102.4
104.4
106.4
110.4
111.0
110.7
112.2
121.0
107.5
104.3
102.1
104.2
106.2
110.3
111.1
110.7
112.2
121.0
1980 .......................
1981 .......................
1982 .......................
1983 .......................
1984 .......................
1985 .......................
1986 .......................
1987 .......................
1988 .......................
1989 .......................
152.1
156.6
156.6
154.0
157.8
159.3
163.0
164.0
168.7
174.7
155.6
160.1
160.2
157.3
160.7
162.0
165.5
166.0
170.4
176.6
125.8
126.4
121.7
119.0
124.9
123.0
121.2
119.5
118.4
120.2
125.8
126.4
121.8
119.2
125.1
123.1
121.4
119.7
118.3
120.1
1990 .......................
1991 .......................
1992 .......................
175.4
173.0
171.2
177.0
174.0
172.1
116.7
112.2
113.7
116.1
111.3
113.4
Average annual
rate of change:
1967-92 ..................
2.2
2.2
0.5
0.5
Year
Enterprise Service—71
Labor inputs
There were about 206,000 State and local government employees in the transit field
in 1992, 21,000 State and 185,000 local government. Special districts (60 percent),
municipalities (35 percent), and counties (5 percent) employed the local government
workers.82
Most (93 percent) transit employees are full-time employees. Comparable figures
for electric, gas, and water are 97, 96 and 92, respectively. There has been a slight, but
steady decrease in the percent of full-time transit employees over the past 25 years.
Census statistics show that 99.5 percent were full-time employees in 1967; 98.7 percent
in 1977; 95.0 percent in 1987; and 93.4 percent in 1992.
Two labor indexes, full-time-equivalent and total employment, are normally calculated for government productivity measures. The two indexes often move in concert,
which has been the case with the enterprise services examined thus far, and is the case
for mass transit. Because they do track closely, and data are readily available with
which to calculate an FTE index, the calculations in this bulletin will use full-timeequivalent employees.
The employment indexes presented here were constructed from individual transit
system data. It was necessary to collect and assemble the data by individual system to
ensure that the data base included only government systems. Two separate periods and
several different data sets were used in constructing the index. For 1967-78, data were
taken primarily from APTA Transit Operating Reports, individual transit system reports, and selected research studies. For 1979-92, they were taken primarily from Department of Transportation National Transit Database.
The overall average annual increase in the transit FTE index is 2.6 percent, with
increases in employment in 19 of the 25 years. However, over the last 3 years, 1990-92,
employment decreased each year. The overall sample index parallels the Census transit
index from 1967-87. Since 1987, Census employment has grown while the sample
constructed for this study has decreased. The APTA employment index, which includes private as well as government, has also decreased during this period. This lends
some credence to the sample constructed for this study. Nevertheless, the reason for the
divergence between these indexes is unknown, but may become clearer with the publication of the 1992 Census benchmark data.
Transit modal employment shows very different rates of change. Bus, heavy rail,
and light rail increased while trolley dropped slightly. The average annual increase for
bus was 3.3 percent and 1.8 percent for heavy rail (subway). These two modes account
for over 90 percent of sample transit employment. The largest modal labor increase
was in the light rail (streetcar) index, which rose at an average annual rate of 5.3 percent. Trolley employment dropped by 0.8 percent annually during this period. All
statistics are for 1967-92 (table 32).
Enterprise Service—72
Table 32. Labor indexes for local government mass transit by mode,
1967-92
(1967=100)
National productivity
indexes
Year
Total
Bus
Heavy
rail
Light
rail
Trolley
1967 ........................
1968 ........................
1969 ........................
100.0
102.0
104.8
100.0
100.7
104.3
100.0
104.2
106.3
100.0
101.6
102.4
100.0
94.5
90.2
1970 ........................
1971 ........................
1972 ........................
1973 ........................
1974 ........................
1975 ........................
1976 ........................
1977 ........................
1978 ........................
1979 ........................
117.3
123.3
127.2
135.1
143.3
150.8
156.2
154.8
156.7
159.3
121.1
125.3
130.1
147.7
160.8
174.6
185.0
186.5
189.4
196.2
112.9
121.0
124.0
121.8
125.1
125.0
124.7
119.7
118.4
115.8
170.2
186.1
189.9
191.9
203.1
219.4
220.3
226.7
298.1
301.7
88.1
89.8
89.8
74.1
63.2
66.4
66.1
64.5
65.1
68.4
1980 ........................
1981 ........................
1982 ........................
1983 ........................
1984 ........................
1985 ........................
1986 ........................
1987 ........................
1988 ........................
1989 ........................
169.2
174.4
177.5
177.0
182.1
190.3
193.3
197.1
196.7
200.7
207.4
211.5
215.0
212.8
218.3
227.3
228.2
233.6
233.4
239.8
124.7
131.4
134.0
134.7
140.3
148.8
152.2
154.1
153.4
154.7
302.3
303.8
306.3
308.7
292.4
271.0
329.7
349.8
360.1
362.0
73.0
76.2
81.8
92.6
92.5
91.0
101.7
99.2
95.4
94.6
1990 ........................
1991 ........................
1992 ........................
198.2
197.0
192.5
236.4
231.1
225.4
152.8
157.2
154.4
372.5
382.5
363.7
90.4
85.8
81.3
Average annual
rate of change:
1967-92 ...................
2.6
3.3
1.8
5.3
-0.8
Local government transit labor and output increased during the 1967-92 period, but
labor increased at a much more rapid rate, the net result being decreasing labor productivity. Although VRM and UPT both dropped during the period, unlinked passenger
trips decreased at a somewhat greater rate (2.1 versus 0.5) as the number of vehicle
miles driven has increased much more rapidly than the number of passengers using
mass transit. Unlinked passenger trip labor productivity decreased in 20 of the 25
measured years, and continues to decrease. Vehicle revenue miles, however, have remained relatively stable, at least since 1982 (table 33).
Enterprise Service—73
Table 33. Two labor productivity indexes for local
government mass transit service, 1967-92
(1967=100)
Year
Vehicle
revenue
miles
Unlinked
passenger
trips
1967 .............................
1968 .............................
1969 .............................
100.0
99.7
99.5
100.0
98.3
96.5
1970 .............................
1971 .............................
1972 .............................
1973 .............................
1974 .............................
1975 .............................
1976 .............................
1977 .............................
1978 .............................
1979 .............................
101.1
98.5
96.9
97.6
94.9
94.3
94.6
94.8
93.4
92.7
91.9
84.8
80.5
77.3
74.2
73.2
71.1
71.5
71.6
76.0
1980 .............................
1981 .............................
1982 .............................
1983 .............................
1984 .............................
1985 .............................
1986 .............................
1987 .............................
1988 .............................
1989 .............................
89.9
89.8
88.2
87.0
86.7
83.7
84.4
83.2
85.8
87.0
74.3
72.5
68.5
67.2
68.6
64.6
62.7
60.6
60.2
59.9
1990 .............................
1991 .............................
1992 .............................
88.5
87.8
88.9
58.9
56.9
59.1
Average annual
rate of change:
1967-92 ........................
-0.5
-2.1
Modal trend comparisons
Much the same picture—increasing employment and output and decreasing labor
productivity—is found in the individual modal comparisons. Statistics for bus and
heavy rail are presented here. Bus, which dominates mass transit, shows the same
general trends as overall transit—increasing employment and output, and decreasing
productivity. The drop in unlinked passenger trip productivity is much more rapid than
the decrease in vehicle revenue mile productivity. Heavy rail, the second most important transit mode, also shows dropping labor productivity for both outputs (tables 34
and 35).
Enterprise Service—74
Table 34. Productivity indexes for local government bus service, 1967-92
(1967=100)
Year
FTE
employees
VRM
output
UPT
output
VRM
UPT
productivity productivity
1967 .................................
1968 .................................
1969 .................................
100.0
100.7
104.3
100.0
101.3
103.0
100.0
100.6
101.1
100.0
100.6
98.7
100.0
100.0
96.9
1970 .................................
1971 .................................
1972 .................................
1973 .................................
1974 .................................
1975 .................................
1976 .................................
1977 .................................
1978 .................................
1979 .................................
121.1
125.3
130.1
147.7
160.8
174.6
185.0
186.5
189.4
196.2
120.9
124.3
129.6
151.9
159.2
175.0
188.7
192.6
192.7
193.3
108.8
107.4
106.3
113.7
118.6
127.0
130.6
130.2
130.4
143.6
99.8
99.2
99.6
102.9
99.0
100.2
102.0
103.3
101.8
98.5
89.8
85.7
81.7
77.0
73.7
72.7
70.6
69.8
68.9
73.2
1980 .................................
1981 .................................
1982 .................................
1983 .................................
1984 .................................
1985 .................................
1986 .................................
1987 .................................
1988 .................................
1989 .................................
207.4
211.5
215.0
212.8
218.3
227.3
228.2
233.6
233.4
239.8
197.4
202.9
203.2
197.4
198.4
197.5
200.0
198.6
201.0
209.3
150.6
152.1
144.2
136.5
145.2
140.0
135.6
130.5
130.9
132.6
95.1
95.9
94.5
92.8
90.9
86.9
87.6
85.0
86.1
87.3
72.6
71.9
67.1
64.2
66.5
61.6
59.4
55.9
56.1
55.3
1990 .................................
1991 .................................
1992 .................................
236.4
231.1
225.4
208.7
204.8
201.7
131.1
127.5
124.6
88.3
88.6
89.5
55.4
55.2
55.3
Average annual
rate of change:
1967-92 ............................
3.3
2.8
0.9
-0.4
-2.3
Enterprise Service—75
Table 35. Productivity indexes for local government heavy rail service, 1967-92
(1967=100)
Year
FTE
employees
VRM
output
UPT
output
VRM
UPT
productivity productivity
1967 ...............................
1968 ...............................
1969 ...............................
100.0
104.2
106.3
100.0
102.7
107.2
100.0
100.5
102.1
100.0
98.6
100.9
100.0
96.5
96.1
1970 ...............................
1971 ...............................
1972 ...............................
1973 ...............................
1974 ...............................
1975 ...............................
1976 ...............................
1977 ...............................
1978 ...............................
1979 ...............................
112.9
121.0
124.0
121.8
125.1
125.0
124.7
119.7
118.4
115.8
114.7
116.5
113.7
108.8
110.1
106.2
103.1
96.8
95.4
97.6
105.0
99.7
96.0
92.1
90.8
90.0
87.1
86.6
89.4
93.1
101.6
96.3
91.7
89.3
88.0
85.0
82.7
80.8
80.6
84.2
93.0
82.4
77.4
75.6
72.6
72.0
69.8
72.3
75.5
80.4
1980 ...............................
1981 ...............................
1982 ...............................
1983 ...............................
1984 ...............................
1985 ...............................
1986 ...............................
1987 ...............................
1988 ...............................
1989 ...............................
124.7
131.4
134.0
134.7
140.3
148.8
152.2
154.1
153.4
154.7
102.5
105.9
105.6
106.4
113.5
117.8
123.1
126.2
133.2
137.0
93.6
93.0
91.0
94.3
97.1
100.1
102.1
104.4
100.3
102.2
82.2
80.6
78.8
79.0
80.9
79.2
80.9
81.9
86.9
88.5
75.1
70.8
67.9
70.0
69.2
67.3
67.0
67.8
65.4
66.0
1990 ...............................
1991 ...............................
1992 ...............................
152.8
157.2
154.4
138.2
135.7
135.2
94.3
87.3
95.9
90.4
86.3
87.6
61.7
55.5
62.1
Average annual
rate of change:
1967-92 ..........................
1.8
1.2
-0.2
-0.5
-1.9
System trend comparisons
Thus far the discussion has focused on national trends. Individual system trends,
which are of particular interest to facility operators, can also be calculated. This section presents productivity statistics for four of the six largest systems: New York, Chicago, Los Angeles, and Boston. Philadelphia and Washington, the other two systems
that comprise the big six, are excluded because they were not included in the data base
until the 1970s. The four large systems accounted for about 40 percent of all local
government transit passenger trips in 1992.
The New York Metropolitan Transit Authority is the largest transit system in the
United States by far. In 1992 it produced one-quarter of all unlinked passenger trips, in
1967 about one-half. New York also dominates transit employment. In 1967 it accounted for 54 percent of all employment, in 1992, 25 percent. Thus, it is hardly surprissing that the New York indexes mirror the national indexes.
For the entire period, 1967-92, the New York index of unlinked passenger trip productivity decreased 2.2 percent per year while the overall national index dropped 2.1
percent (table 36). For vehicle revenue mile productivity the figures were -0.3 percent
and -0.5 percent, respectively (table 36 and 37).
Enterprise Service—76
Table 36. Comparison of unlinked passenger trip productivity indexes for total
sample and New York City, Chicago, Los Angeles, and Boston transit systems,
1967-92
(1967=100)
Year
Total
New York
Chicago
Los Angeles
Boston
1967 ...................
1968 ...................
1969 ...................
100.0
98.3
96.5
100.0
97.5
97.4
100.0
98.3
95.0
100.0
100.8
104.8
100.0
96.6
85.1
1970 ...................
1971 ...................
1972 ...................
1973 ...................
1974 ...................
1975 ...................
1976 ...................
1977 ...................
1978 ...................
1979 ...................
91.9
84.8
80.5
77.3
74.2
73.2
71.1
71.5
71.6
76.0
93.5
83.3
79.1
77.2
75.3
75.1
70.7
73.7
76.1
78.3
93.2
87.6
87.1
84.5
82.2
79.5
82.1
86.9
75.3
94.8
103.2
99.6
90.8
92.5
83.8
72.3
83.0
77.8
84.5
104.8
84.3
74.4
74.6
80.2
79.6
84.9
85.5
94.4
73.7
74.5
1980 ...................
1981 ...................
1982 ...................
1983 ...................
1984 ...................
1985 ...................
1986 ...................
1987 ...................
1988 ...................
1989 ...................
74.3
72.5
68.5
67.2
68.6
64.6
62.7
60.6
60.2
59.9
76.6
73.8
69.5
67.8
66.2
63.4
63.8
63.1
60.8
68.7
93.4
90.3
86.8
89.9
93.2
99.0
94.6
87.8
87.7
88.1
106.0
98.9
88.8
100.0
105.3
112.8
93.7
86.5
82.6
104.9
63.1
67.4
64.8
75.9
76.3
79.3
87.7
87.8
91.0
86.9
1990 ...................
1991 ...................
1992 ...................
58.9
56.9
59.1
62.5
56.7
58.0
86.3
80.0
77.9
101.4
106.8
101.1
92.1
88.5
100.1
Average annual
rate of change:
1967-92 ..............
1967-90 ..............
-2.1
-2.3
-2.2
-2.0
-1.0
-.6
0.0
.1
0.0
-.4
1982-92 ..............
1982-90 ..............
-1.5
-1.9
-1.8
-1.3
-1.1
-.1
1.3
1.7
4.4
4.5
Enterprise Service—77
Table 37. Comparison of vehicle revenue mile productivity indexes for total sample
and New York City, Chicago, Los Angeles, and Boston transit systems, 1967-92
(1967=100)
Year
Enterprise Service—78
Total
New York
Chicago
Los Angeles
Boston
1967 ...................
1968 ...................
1969 ...................
100.0
99.7
99.5
100.0
99.3
100.7
100.0
99.2
95.8
100.0
101.9
103.8
100.0
98.6
98.3
1970 ...................
1971 ...................
1972 ...................
1973 ...................
1974 ...................
1975 ...................
1976 ...................
1977 ...................
1978 ...................
1979 ...................
101.1
98.5
96.9
97.6
94.9
94.3
94.6
94.8
93.4
92.7
100.2
96.7
93.9
91.7
88.7
86.3
83.0
84.1
82.3
81.8
94.7
91.8
91.1
89.7
87.7
84.7
83.0
84.0
81.2
87.2
102.9
102.7
98.1
102.4
94.9
76.7
85.2
87.1
88.4
89.1
87.8
92.4
93.5
96.7
91.7
93.5
97.9
94.8
78.9
74.2
1980 ...................
1981 ...................
1982 ...................
1983 ...................
1984 ...................
1985 ...................
1986 ...................
1987 ...................
1988 ...................
1989 ...................
89.9
89.8
88.2
87.0
86.7
83.7
84.4
83.2
85.8
87.0
84.3
84.3
81.9
81.5
83.8
79.3
81.7
81.3
94.1
95.5
83.7
83.5
82.1
83.5
81.9
87.5
87.3
84.0
85.9
87.5
82.7
81.0
78.4
74.4
72.6
72.0
65.6
63.4
62.4
75.7
76.3
82.8
87.2
95.4
97.3
97.9
112.4
113.4
116.9
118.3
1990 ...................
1991 ...................
1992 ...................
88.5
87.8
88.9
96.4
93.3
93.1
87.3
88.6
86.8
75.6
78.0
73.1
122.1
117.7
133.9
Average annual
rate of change:
1967-92 ..............
1967-90 ..............
-0.5
-.5
-0.3
-.2
-0.6
-.6
-1.2
-1.2
1.2
.9
1982-92 ..............
1982-90 ..............
.1
0
1.3
2.1
.6
.8
-.7
-.4
4.4
4.3
Average annual rates of change for 1982-92 are presented in the tables. Prior to
1980, there are questions concerning data accuracy. Beginning year 1982 is used because it is a base year in the calculations. For 1982-92, the overall national passenger
trip productivity index showed a change of -1.5 percent annually while New York registered -1.8 percent. For the vehicle revenue mile productivity index the national rate
was 0.1 percent annually and the New York rate was 1.3 percent.
New York operates buses and heavy rail. Over the long term (1967-92), the vehicle
revenue mile productivity index decreased for both modes at almost the same annual
rate (-0.3 percent). As a point of comparison, the national statistics were -0.4 percent
for buses and -0.5 percent for heavy rail (tables 34 and 35). The New York passenger
trip annual rate was -3.3 percent for buses and -1.7 percent for heavy rail; the national
statistics were -2.3 percent and -1.9 percent, respectively.
Chicago operated the second largest transit system in the United States in 1992,
although it was a distant second to New York. Like New York, and the national sample,
the Chicago Regional Transportation Authority showed declining output per employee
year statistics for 1967-92. This is true for both the passenger trip (-1.0 percent) and
revenue mile (-0.6 percent) productivity measures. Chicago operates bus and heavy
rail modes, and until 1973 it also operated trolley buses. All three modes are included
in the Chicago index.
Los Angeles (Southern California Rapid Transit District), the third largest system,
operated only buses during the period covered by the indexes; light rail was added in
1990, but it is not included here. While most national and individual transit system
productivity indexes show passenger trip productivity lagging vehicle revenue mile
productivity, Los Angeles shows the opposite. The average annual rate of change for
passenger trip productivity was essentially flat from 1967-92 while vehicle revenue
mile productivity decreased at 1.2 percent per year.
Boston (Massachusetts Bay Transportation Authority) operated four modes in 1992:
Bus, heavy rail, light rail, and trolley. It is the only system of the four that shows longterm vehicle revenue mile productivity gains (table 37). These gains have been particularly pronounced since the early 1980s. Moreover, each mode registered vehicle revenue mile productivity gains over the long run and since 1982.
Boston’s 1992 unlinked passenger trip productivity index stood at about where it did
in 1967 although there was a decline in the intervening years. In the case of the individual modes, the bus and trolley indexes showed decreasing labor productivity over
the long term (1967-92) while heavy rail and light rail increased. All four modes registered increasing productivity between 1982 and 1992.
Modal level comparisons
Productivity levels, in contrast to trends, can provide a very different perspective or
dimension on productivity. The conceptual and data requirements to calculate levels
are much more demanding than those needed to calculate trends, and for most government services it is infeasible to calculate meaningful national productivity levels. Transit is an exception. As noted above, there is considerable research and general agreement as to what should be calculated, and good data are available on the transit industry.
Data for 1992 show that, on the average, there were 10,770 vehicle revenue miles
logged for each full-time-equivalent operational employee in the sample. The statistics
for the individual modes are the following: Bus, 10,930 miles; heavy rail, 10,840 miles;
trolley, 7,930 miles; and light rail, 6,520 miles.
These data are available by system. The following array presents data for nine bus
systems in 1992. These systems were selected for purposes of illustration only. That is,
mid-size systems are included that have about the same number of vehicle revenue
miles per FTE employee. Vehicle revenue mile and unlinked passenger trip per FTE
Enterprise Service—79
operational employee for the nine systems and the overall sample average show the
following:
System
All systems
Sacramento, CA
Long Beach, CA
Norfolk, VA
Santa Clara, CA
Alameda, CA
Tacoma, WA
Dade County, FL
St. Louis, MO
Atlanta, GA
VRM/FTE
10,930
12,840
12,760
12,720
12,300
12,260
12,230
12,210
12,170
12,160
UPT/FTE
38,800
29,820
40,260
18,210
28,420
35,570
21,510
32,930
28,450
36,910
It is interesting to note that those bus systems which operate in the congested Eastern
cities such as Boston, Philadelphia, and New York, show much lower vehicle revenue
miles per employee. New York City, for example, registered only 6,680 vehicle miles
per FTE employee. However, the picture is very different for unlinked passenger trips.
The average number of passengers carried per FTE employee for all sample systems is
38,800 for 1992. But New York registered 48,320 and Chicago 54,230.
Most of the systems shown in the above array fall below the UPT sample average,
and some, such as Norfolk, are far below the norm. However, Long Beach rises above
the average for both passenger trips and revenue miles per employee.
State Alcoholic
Beverage Sales
All States and many local governments regulate alcoholic beverage sales. States
license sellers, tax sales, regulate advertising, set the legal age for purchase of beverages, and establish the hours and days of sale. In addition, about one-third of the States
operate wholesale warehouses and/or retail alcoholic beverage stores. These operations are the focus of this section.83
Institutional setting
States which operate their own warehouses and retail stores using government employees are known as control or monopoly States. Those that use private sellers are
known as license States. There are 18 control States and 33 license States including the
District of Columbia in the United States. This discussion focuses on the control States.
(table 38).
Although States are often divided between the issue of control and license, there is a
broad spectrum of institutional arrangements in this area, from almost completely private operation to almost total government control and operation. These arrangements
can be grouped into the following fairly distinct categories:
•
•
•
•
•
•
Private retail and wholesale operations, in effect in more than half of the
States.
Private retail and government wholesale, as in Iowa, Mississippi, and
Wyoming.
Private and government (municipal) retail and private wholesale, as in
Minnesota.
Government (city and county) retail and private wholesale, as in North
Carolina.
Government and private agency retail and government wholesale, as in
Ohio.
Government retail and wholesale, as in Alabama and Virginia.
This study focuses on those States included under the second, fifth, and sixth category above. North Carolina is not included, because local authorities rather than State
personnel operate the stores, and local data are not readily available with which to make
productivity calculations.
Enterprise Service—80
The control States accounted for about 28 percent of the U.S. population and 24
percent of the gallons of spirits sold in the United States in 1992.84 In that same year,
State and local government alcoholic beverage sales totaled over $3.6 billion, with
State government alone accounting for $3.1 billion.
Five control States—Pennsylvania, Michigan, Ohio, Virginia, and Washington—
accounted for 66 percent of the revenue and 73 percent of the employment in fiscal
1992. Pennsylvania alone accounted for 22 percent of the revenue and 31 percent of
the employment. These percentages have remained fairly stable over the past 20 years.85
Most State alcoholic beverage control commissions are responsible for four functions: Wholesale sales, retail sales, licensing, and enforcement. Each of the 17 control
States operate wholesale alcoholic beverage facilities. Sixteen of them license others,
such as wineries and restaurants, to sell alcohol. Twelve of the 17 are responsible for
enforcement of State alcoholic beverage laws and regulations such as after hour sales
and sales to minors. The other control States assign enforcement to State police, Departments of Public Safety, or in the case of Wyoming, to local government (table 38).
Enterprise Service—81
Table 38. Type of State alcoholic beverage control distribution, 1992
Type of sales
State
Wholesale
State
Retail
State
Retail
agent
Alabama .............
Idaho ..................
Iowa ....................
Maine .................
Michigan .............
X
X
X
X
X
X
X
—
X
—
—
X
—
—
—
Mississippi ..........
Montana .............
New Hampshire .
Ohio ...................
Oregon ...............
X
X
X
X
X
—
X
X
X
—
—
X
X
X
X
Pennsylvania ......
Utah ...................
Vermont ..............
Virginia ...............
Washington ........
X
X
X
X
X
X
X
X
X
X
—
—
X
—
X
West Virginia ......
Wyoming ............
X
X
—
—
—
—
Type of beverage sold by State
State
Spirits
Wine
Beer
Alabama .............
Idaho ..................
Iowa ....................
Maine .................
Michigan .............
X
X
X
X
X
X
X
—
X
—
—
X
—
—
—
Mississippi ..........
Montana .............
New Hampshire .
Ohio ...................
Oregon ...............
X
X
X
X
X
X
X
X
—
—
—
—
—
—
—
Pennsylvania ......
Utah ...................
Vermont ..............
Virginia ...............
Washington ........
X
X
X
X
X
X
X
X
X
X
—
X
—
—
X
West Virginia ......
Wyoming ............
X
X
X
X
—
—
SOURCE: Personal communication from Jim Squeo of National Association
of Beverage Control Authorities, October 1994, and Summary of State Laws and Regulations Relating to Distilled Spirits, Washington: Distilled Spirits Council of the United
States, December, 1993.
This investigation includes wholesale as well as retail sales. Six of the 17 States are
wholesale-only operations; the other 11 operate a combination of wholesale and retail.
All of the 11 State retail operations sell spirits; 10 of these sell wine as well as spirits,
and 3 also sell beer. Spirits account for about 80 percent of all gallons sold in State
stores, wine about 20 percent, and beer less than 0.01 percent. Six of the 11 States that
operate retail stores use agency (private) stores to augment their operations.
The agency arrangement is an important issue in calculating productivity. This arises
when private retail merchants contract to operate agency outlets as part of their normal
Enterprise Service—82
operations. Control States have long used agents to serve sparsely populated areas
where a “full-service” State store could not be justified. Ohio, for example, permits
agents in municipalities with a population of less than 20,000. Agents are usually paid
a percent of their gross sales, although some States pay a fixed fee or negotiate a price
with the individual merchant. In all cases, prices of alcoholic beverages are set by the
State.
In recent years States have increasingly turned to agent sales as they desired to cut
back on government employment. Oregon, for example, reduced the number of its
State stores from 20 in 1976 to 6 in 1980 before it turned entirely to agent sales in 1983.
Maine, Montana, and Utah, among others, have also substituted agents for State-operated stores.
Utah uses three different forms of agency arrangements. In the first, agents operate
beverage stores just as they might a State store, but do not hire State employees. In the
second, merchants contract to sell alcoholic beverages in addition to their regular product lines. In the third, resort or hotel owners operate an agency store as a convenience
to their guests, usually at no cost to the State.
In addition to the 17 control States, 6 States permit or require local government
sales. North Carolina requires local government sales if alcoholic beverages are sold
for use off-premise. Alaska, Maryland, Minnesota, South Dakota, and Wisconsin permit local option. Minnesota has a combination of private and municipal liquor stores.
Alcoholic beverage retail sales fall under SIC code 5921. The SIC defines alcoholic
beverage retail stores as:
Establishments primarily engaged in the retail sale of packaged
alcoholic beverages, such as ale, beer, wine, and liquor, for consumption off the premises.86
Wholesale spirit and wine sales come under code 5182 and are defined as:
Establishments primarily engaged in the wholesale distribution of
distilled spirits, including neutral spirits and ethyl alcohol used in
blended wines and distilled liquors.87
Alcoholic beverage control boards are assigned to SIC 9651, regulation, licensing, and
inspection of commercial sectors.
The Bureau of Census, which generates much of the data used here, has two slightly
different definitions for State alcoholic beverage operations. For financial transactions,
it defines a “liquor store” as an alcoholic beverage distribution facility “operated by
governments maintaining alcoholic beverage monopoly systems.” It excludes expenditure for law enforcement and licensing activities carried out in conjunction with liquor
store operations.88
For employment statistics, Census limits its definition to the “administration and
operation of retail liquor stores operated by State governments.”89 However, statistics
collected under this heading include wholesale as well as retail operations, and some
licensing and enforcement personnel. The Census employment statistics do not include
any local government personnel.
Outputs
This section discusses the measurement of State alcoholic beverage store outputs,
and presents several indexes for such measurements. As noted earlier, many alcoholic
beverage control authorities are responsible for enforcement and licensing as well as
sales. Enforcement and licensing operations, which account for less than 10 percent of
beverage commission employees, require a different set of output measures. They are
not considered in this review, which focuses on sales of alcoholic beverages.
Sales of alcoholic beverages are commonly measured in one of five ways—dollars,
customers, bottles, cases, or gallons.90 Probably the best measure of output for calcuEnterprise Service—83
lating labor productivity is the number of bottles sold, in the case of retail sales, and the
number of cases sold for wholesale sales. These two measures reflect a large part of the
physical work involved in alcoholic beverage sales. However, data to calculate these
two measures are not available for the early years.
The output measure that is used here is the number of gallons sold. It is physical,
measurable, repetitive, easily understood, and the final output. Also, it is a readily
available statistic that correlates highly with bottles and cases.
Although gallons are used as the basic measure of output, they are divided into three
different services for productivity measurement purposes. First, is retail sales by government employees in State stores. Second, is retail sales by agents in private stores.
These sales are handled by private employees but State employees handle wholesale
and audit functions. The third service is wholesale sales by State employees; in this
case all retail sales are by the private sector. The reasons that the three services are
measured separately is that each has a different output, the outputs seem to change at
very different rates, and two of the three outputs have very different unit labor requirements.
The index numbers for the three types of sales are shown in table 39. State retail
store sales, as measured by the number of gallons sold, increased until 1979, at which
point they start to drop. The average annual increase over the 1967-79 period was 2.7
percent. This reflects growing population, increases in disposable income, increases in
the number of State stores, and promotion of sales in a number of States. In 1979 there
was an abrupt change in industry growth and from 1979-92 sales decreased at a rate of
3.1 percent per annum. The decrease reflected a number of factors including decreasing per capita consumption, a shift by the States from retail to wholesale only, and to
agent operations.
Agent sales are a small, but growing part of alcoholic beverage sales as more States
turn to agents to handle their sales, particularly in the less densely populated areas. The
long-term (1967-92), average annual agent sales increased by 4.1 percent. The average
annual increase between 1967 and 1979 was 7.7 percent, but between 1979 and 1992 it
had declined to 1.0 percent. There was a dramatic increase in 1992 when Ohio transferred a large part of its operations from State stores to agent operations.
There has been a modest increase in State wholesale sales during the measured period as States moved from retail to wholesale-only operations. The long-term average
annual State wholesale sales increased by 1.4 percent. In 1967 there were three wholesale-only States, Michigan, Mississippi, and Wyoming, but by 1992 the number increased to five with the addition of Iowa and West Virginia. Even with the increase in
the number of States there has been a slowdown in wholesale sales in recent years.
Between 1979 and 1992, State government wholesale sales dropped, on average, 0.7
percent annually.
Table 39 also shows the overall change in output for State alcoholic beverage sales,
both unweighted and weighted. The unweighted index reflects the total number of
gallons sold by the States through the State-operated stores, agents, and warehouses.
This index shows sales increasing at an average annual rate of 3.2 percent between
1967 and 1979 with increases in every year. Between 1979 and 1992, sales dropped in
most years and the average annual rate of change was -2.2 percent.
The weighted index, which conceptually is the preferred index, reflects the retail,
agent, and wholesale indexes weighted by their appropriate unit labor weights. To
calculate the overall index, individual 5-year segments are calculated, each segment is
base year weighted (1967, 1972, 1977, 1982, and 1987), and the individual segments
are then linked to form the index.91 The results of these calculations show sales increasing at an average annual rate of 2.9 percent between 1967 and 1979 with each year
registering an increase. From 1979 through 1992, sales declined at an average annual
rate of 2.8 percent, and output dropped every year. This rather abrupt change in output
reflects the decrease in alcoholic beverage sales nationally and the shift from State to
private sector operations. Over the long term (1967-92), the average annual rate of
change is essentially flat.
Enterprise Service—84
Table 39. State alcoholic beverage sale indexes, 1967-92
(1967 = 100)
Wholesale
index
Unweighted
output
index
Weighted
output
index
Year
Retail
index
Agent
index
1967 .......................
1968 .......................
1969 .......................
100.0
103.5
108.1
100.0
109.9
124.5
100.0
107.0
113.9
100.0
104.4
109.8
100.0
103.9
108.8
1970 .......................
1971 .......................
1972 .......................
1973 .......................
1974 .......................
1975 .......................
1976 .......................
1977 .......................
1978 .......................
1979 .......................
111.3
116.0
120.2
126.3
126.4
131.2
132.2
133.7
137.6
138.0
128.2
136.0
146.1
161.2
172.8
185.8
195.1
204.2
220.2
242.2
119.9
123.5
133.0
136.0
139.5
141.4
143.8
148.1
154.5
154.8
113.6
118.1
123.6
129.4
130.6
135.2
136.8
139.1
143.9
145.1
112.2
116.8
121.5
127.6
128.0
132.8
133.9
135.7
139.9
140.5
1980 .......................
1981 .......................
1982 .......................
1983 .......................
1984 .......................
1985 .......................
1986 .......................
1987 .......................
1988 .......................
1989 .......................
136.9
135.4
131.6
127.2
123.7
119.2
115.3
108.0
102.3
99.6
249.7
261.0
256.1
253.6
248.2
241.5
239.0
237.2
232.8
227.3
155.0
151.6
148.8
141.3
140.8
138.2
131.8
141.3
143.1
138.6
144.6
143.2
139.6
134.7
131.6
127.4
123.1
119.1
114.9
111.7
139.6
138.2
134.4
129.9
126.4
122.1
118.0
111.7
106.5
103.6
1990 .......................
1991 .......................
1992 .......................
98.3
92.7
91.9
233.1
235.1
275.9
143.1
141.6
141.2
111.8
107.3
108.1
102.7
97.6
97.3
Average annual
rate of change:
1967-92 ..................
1967-79 ..................
1979-92 ..................
-0.3
2.7
-3.1
4.1
7.7
1.0
1.4
3.7
-.7
0.3
3.2
-2.2
-0.1
2.9
-2.8
Labor inputs
The number of employees and the number of full-time equivalent employees are the
two input measures suggested in chapter 2 for calculating State and local government
labor indexes. When there are few part-time employees or little overtime work, as was
the case with electric power, gas, water supply, and transit, the two labor series yield
very similar trends. However, when there are a number of part-time employees and the
ratio between part time and full time is changing, or the overtime worked changes through
time, then the two indexes can diverge.
In the case of State alcoholic beverage operations, most statistics point to a change
in the ratio between full-time and part-time employment. Furthermore, State retail stores
use a large number of part-time employees. According to the Bureau of Census, parttime employment accounted for 17 percent of all State alcoholic beverage employees in
1977, but by 1992 the number had increased to 31 percent. Total employment data are
not readily available prior to 1977 so it isn’t known when the trend to increased use of
part-time employees started. Also, the lack of data precludes the calculation of an
index of the total number of employees.
A further potential complication in calculating a labor index, is the large number of
part-time and intermittent beverage sales workers hired in the December holiday season. In addition, full-time employees often work overtime at this time. None of the
Enterprise Service—85
data reflect increased holiday employment because all the employment data are for the
month of October. The calculations presented here assume that the ratio between October and the Holiday season has not changed.
Another potential problem in calculating an alcoholic beverage labor series is the
inclusion of State enforcement and licensing personnel (and drug enforcement personnel in one State) in the summary statistics. Output statistics do not measure enforcement and licensing activities, and any employment index should exclude the personnel
who perform these activities. However, the State data do not separately identify these
individuals. Nevertheless, it is possible to estimate these individuals using State and
National Association of Beverage Control Authorities (NABCA) data, and remove them
from the overall totals.
The State alcoholic beverage sales FTE employment index has declined steadily
over time (table 40). The long-term (1967-92) average annual employment decrease
was 1.2 percent with employment dropping in 17 of the 25 years. Since 1979, employment has decreased every year except for 1985.
Table 40. State alcoholic beverage control sale productivity indexes,
1967-92
(1967 = 100)
Output
index
Labor
index
Productivity
index
1967 .................................
1968 .................................
1969 .................................
100.0
103.9
108.8
100.0
103.1
106.7
100.0
100.8
101.9
1970 .................................
1971 .................................
1972 .................................
1973 .................................
1974 .................................
1975 .................................
1976 .................................
1977 .................................
1978 .................................
1979 .................................
112.2
116.8
121.5
127.6
128.0
132.8
133.9
135.7
139.9
140.5
111.2
109.2
109.7
108.1
107.6
108.2
108.3
106.2
107.3
105.9
100.9
107.0
110.8
118.0
119.0
122.8
123.7
127.8
130.3
132.8
1980 .................................
1981 .................................
1982 .................................
1983 .................................
1984 .................................
1985 .................................
1986 .................................
1987 .................................
1988 .................................
1989 .................................
139.6
138.2
134.4
129.9
126.4
122.1
118.0
111.7
106.5
103.6
105.8
104.8
99.7
94.5
91.2
93.1
92.7
84.5
82.3
80.2
131.9
131.9
134.8
137.4
138.7
131.1
127.4
132.2
129.4
129.3
1990 .................................
1991 .................................
1992 .................................
102.7
97.6
97.3
79.2
75.3
73.2
129.8
129.6
132.8
Average annual
rate of change:
1967-92 ............................
-0.1
-1.2
1.1
1967-79 ............................
1979-92 ............................
2.9
-2.8
.5
-2.8
2.4
0
Year
SOURCE: Table 39 and computed
Enterprise Service—86
Productivity indexes
The drop in employment coupled with relatively stable output has resulted in longterm increasing labor productivity for State alcoholic beverage operations. Between
1967 and 1992, the average annual increase was 1.1 percent. However, there have been
two fairly distinct periods of change. During the period 1967-79, productivity increased,
on average, by 2.4 percent annually; in later years, there has been little or no change
(table 40).
Some States calculate productivity levels, such as the number of bottles or gallons
sold per employee, for the State as a whole, and for individual stores. Such information
can be extremely helpful in managing operations. However, because the type of service
and organizational structure varies from State to State, productivity levels have not
been calculated.
Government-private
comparisons
Comparison of State and private alcoholic beverage sale labor productivity indexes
show that between 1972 (the first year for which data for the private sector are available) and 1992, the average annual increase for the private sector and State government
was the same, 0.9 percent. However, the rate of increase varied by the time period
examined. The major growth period for State government productivity was in the early
years, for the private sector it was the later years (table 41).
Table 41. Comparison of State government and private sector
alcoholic beverage labor productivity indexes, 1972-92
(1972 = 100)
Year
State
government
index
Private
sector
index
1972 ......................................
1973 ......................................
1974 ......................................
1975 ......................................
1976 ......................................
1977 ......................................
1978 ......................................
1979 ......................................
100.0
106.4
107.3
110.7
111.6
115.3
117.6
119.8
100.0
100.0
98.6
95.5
100.0
99.3
94.2
95.6
1980 ......................................
1981 ......................................
1982 ......................................
1983 ......................................
1984 ......................................
1985 ......................................
1986 ......................................
1987 ......................................
1988 ......................................
1989 ......................................
119.0
119.0
121.6
124.0
125.1
118.3
114.9
119.2
116.7
116.7
101.2
103.0
107.2
101.3
99.0
108.0
99.6
106.3
105.3
109.2
1990 ......................................
1991 ......................................
1992 ......................................
117.1
116.9
119.8
114.6
116.0
120.1
Average annual
rate of change:
1972-92 .................................
0.9
0.9
1972-77 .................................
1977-82 .................................
1982-87 .................................
1987-92 .................................
2.9
1.1
-.4
.1
-.1
1.6
-.2
2.5
SOURCE: Table 40 and Productivity Measures for Selected Industries
and Government Services, BLS Bulletin 2440, table 173
Enterprise Service—87
State government output and labor input dropped at a more rapid rate than did
private sector operations. This is partly due to the shift from government to private
sales. The average annual change in State output between 1972 and 1992 was -1.1
percent while private store sales decreased 0.5 percent. Government labor decreased
2.0 percent annually between 1972-90. The decrease for private operations during
this period was 1.4 percent.
These comparisons should be interpreted with care, for several reasons. First, State
government outputs are measured in physical quantities, as discussed above, and private sector outputs are derived by deflated value.92 Second, most private stores stock
and sell non-alcoholic items, and these outputs are measured and included in the private
sector output measure. There are no comparable outputs for State stores. Third, the
private sector is solely a retail sales measure (SIC 5921), but the government measure
includes warehouse or wholesale distribution too (SIC’s 5921 and 5182, respectively).
Nevertheless, it is apparent that many of the changes that have affected private sector operations over the years, including decreasing consumption of spirits by the public,
a shift to self service stores, and computerization of inventory and sales also affected
government operations.
Summary and
Conclusions
This chapter discussed the measurement of State and local government enterprise
services. Five services were examined and average labor productivity indexes calculated for each. Four of these—electric power, natural gas, water supply, and mass transit—compose a group known by the Bureau of Census as State and local government
utilities. The other enterprise service discussed here is State alcoholic beverage sales.
The Bureau of Economic Analysis (BEA) identifies 10 major State and local government functions as enterprise services. They accounted for about 920,000 State and
local government full–time-equivalent employees in 1992 according to the Bureau of
Economic Analysis. The five discussed in this bulletin accounted for 453,000 State and
local government employees. BEA enterprise service employment doubled between
1967-92. Employment among the five services increased about 75 percent (257,000 to
453,000) over the 1967-92 period.
Labor productivity increased for three services and dropped for two over the measured period (table 42). Each service is briefly summarized.
Electric power. Labor productivity increased in this service, on average, 2.1 percent
per year from 1967-92. Output and employment both rose during this period. In 1992,
there were about 85,000 State and local government electric utility employees. Electric
power utility operations is one of the easier services to measure because there is general
agreement concerning the output measure, quality of output is not an issue, and data are
generally available, particularly in recent years.
Natural gas. In contrast to electric power sales, natural gas shows decreasing labor
productivity. The average annual decrease from 1974-92 is 1.6 percent. During the
covered period, output dropped while employment rose. It is a service with a small
number of government employees, fewer than 11,000 in 1992. There is general agreement as to the output measure for this service, and the data have been adequate since
1974.
Water supply. In contrast to electric power and natural gas, water supply is largely a
government owned and operated service. There were about 156,000 local government
water employees in 1992. Local government water utility labor productivity increased
0.6 percent annually between 1967 and 1992. Output and employment also increased
during this period. While there is some question concerning the precise increase in
output and productivity, the numerous statistical series examined for this publication
point to increasing labor productivity. Water supply has one attribute that sets it apart
from the other four services, and that is the importance of quality of service, and how
this has affected output. Drinking water treatment and testing has been expanded dramatically since 1974, and there is general agreement that quality has improved.
Enterprise Service—88
Mass transit. This service shows decreasing labor productivity during the measured
period, 1967-92. When the number of unlinked passenger trips is used as the output
measure, the average annual decrease is 2.1 percent; the decrease for vehicle revenue
hour output is 0.5. Although both measures are useful, the number of passenger trips
seems to be more appropriate as a measure of output for an enterprise service. The data
with which to calculate mass transit labor productivity are reasonably good in the early
years and very good in later years. Mass transit is the largest service examined here in
terms of the number of employees. In 1992 there were about 206,000 State and local
government employees. This service also has seen massive growth in output and employment since 1967, as State and local government took over the operation of most
failed private operations, extending and introducing operations into areas where none
had been provided.
State alcoholic beverage sales. Many changes have occurred in this industry, in both
the private and government sectors over the past 25 years. State output and employment dropped during the 25 years, but employment dropped faster. The result was an
average annual increase in labor productivity of 1.1 percent over the 1967-92 period.
There is general agreement on how outputs should be measured in this service, and the
data are detailed and quite good. Although this is a relatively small State service, with
about 11,000 employees in 1992, it is a function that generates considerable income for
those States that operate retail sales stores.
Comparison of government and private services. Most State and local government
enterprise services have private sector counterparts, including the five discussed here.
Two—mass transit and water supply—are dominated by government, the other three—
electric power, natural gas, and alcoholic beverage sales—are primarily privately owned
and operated. For the latter three services there are private sector labor productivity
indexes. Comparison of government and private labor productivity indexes show remarkably similar trends (table 43, and charts 4 to 6).
For electric power and natural gas, government and private labor productivity trends
show very similar rates of growth. The average annual increase for government operated electric power utilities was 2.1 percent; for the private sector it was 2.3 percent.
For natural gas it was -1.6 and -2.2, respectively. Alcoholic beverage sales show similar long-term trends, but very different periods of growth; State government productivity increased dramatically prior to 1980 while the private sector has shown the greatest
increase after the mid-1980s.
Enterprise services are more easily measured than most other government services.
Most have tangible outputs; are sold in the market place; cover their cost of production
through fees and charges; and are supported by good national data.
Enterprise Service—89
Table 42. Enterprise service labor productivity indexes for five services,
1967-92
(1977 = 100)
Year
Electric
power
1967 ...................
1968 ...................
1969 ...................
Natural
gas
Drinking
water
Mass
transit
(trips)
Mass
transit
(miles)
Alcoholic
beverages
62.9
69.4
78.8
99.2
101.5
99.3
139.9
137.5
134.9
105.5
105.2
105.0
78.2
78.8
79.7
1970 ...................
1971 ...................
1972 ...................
1973 ...................
1974 ...................
1975 ...................
1976 ...................
1977 ...................
1978 ...................
1979 ...................
82.1
85.1
86.6
85.9
85.1
86.6
91.3
100.0
98.2
100.3
107.3
104.1
101.8
100.0
101.1
101.1
98.7
99.1
101.3
101.5
98.5
101.2
100.0
100.0
95.8
106.9
128.6
118.6
112.6
108.1
103.8
102.4
99.5
100.0
100.1
106.3
106.7
104.0
102.2
103.0
100.2
99.6
99.9
100.0
98.5
97.8
79.0
83.7
86.7
92.3
93.1
96.0
96.8
100.0
102.0
103.9
1980 ...................
1981 ...................
1982 ...................
1983 ...................
1984 ...................
1985 ...................
1986 ...................
1987 ...................
1988 ...................
1989 ...................
101.7
99.8
97.4
95.5
94.1
94.5
94.1
96.7
101.5
104.5
107.5
105.7
97.5
96.0
100.3
88.2
79.5
84.2
83.4
81.7
106.2
107.8
104.0
106.3
111.5
118.1
121.2
123.2
122.6
124.9
104.0
101.4
95.9
94.0
95.9
90.4
87.7
84.8
84.2
83.8
94.8
94.7
93.1
91.8
91.5
88.3
89.0
87.8
90.5
91.8
103.2
103.1
105.5
107.5
108.5
102.6
99.7
103.4
101.2
101.2
1990 ...................
1991 ...................
1992 ...................
106.6
107.2
104.5
78.0
80.2
80.6
122.3
114.9
114.5
82.4
79.6
82.6
93.4
92.6
93.8
101.5
101.4
103.9
Average annual
rate of change:
1967-92 ..............
2.1
0.6
-2.1
-0.5
1.1
1974-92 ..............
1977-92 ..............
1.1
.3
-1.6
-1.4
.8
.9
-1.3
-1.3
-.4
-.4
.6
.3
1967-72 ..............
1972-77 ..............
1977-82 ..............
1982-87 ..............
1987-92 ..............
6.6
2.9
-.5
-.1
1.6
-.5
-2.9
-.9
.4
-.3
.8
3.5
-1.5
-4.2
-2.3
-.8
-2.4
-.5
-.6
-.4
-1.4
-1.2
1.3
2.1
2.9
1.1
-.4
.1
SOURCE: Tables 13, 19, 26, 33, and 40
Enterprise Service—90
Table 43. Comparison of labor productivity trends for three government services
and private sector industries, 1967-92
Electric power
Natural gas
Alcohol beverage sale
(1967=100)
(1974=100)
(1972=100)
Year
Government Private Government Private Government Private
1967 ..............
1968 ..............
1969 ..............
100.0
110.3
125.3
100.0
107.6
114.0
1970 ..............
1971 ..............
1972 ..............
1973 ..............
1974 ..............
1975 ..............
1976 ..............
1977 ..............
1978 ..............
1979 ..............
130.6
135.3
137.7
136.6
135.4
137.8
145.3
159.0
156.1
159.5
118.2
124.8
131.5
135.8
133.6
142.4
146.8
153.7
148.6
146.6
100.0
97.0
95.0
93.2
94.3
94.2
1980 ..............
1981 ..............
1982 ..............
1983 ..............
1984 ..............
1985 ..............
1986 ..............
1987 ..............
1988 ..............
1989 ..............
161.7
158.7
154.8
151.8
149.6
150.4
149.7
153.8
161.4
166.2
144.3
143.1
137.3
139.7
145.1
143.5
147.1
154.3
161.9
166.2
1990 ..............
1991 ..............
1992 ..............
169.5
170.5
166.2
169.9
175.0
176.4
2.1
2.3
2.3
2.3
Average annual
rate of change:
1967-92 .........
1967-90 .........
100.0
99.4
101.7
98.1
99.5
101.4
100.0
106.4
107.3
110.7
111.6
115.3
117.6
119.8
100.0
100.0
98.6
95.5
100.0
99.3
94.2
95.6
100.2
98.6
90.9
89.5
93.5
82.2
74.1
78.5
77.7
76.2
100.1
96.3
87.3
79.7
82.0
80.6
72.7
70.6
74.4
73.0
119.0
119.0
121.6
124.0
125.1
118.3
114.9
119.2
116.7
116.7
101.2
103.0
107.2
101.3
99.0
108.0
99.6
106.3
105.3
109.2
72.7
74.8
75.1
66.9
66.2
66.9
117.1
116.9
119.8
114.6
116.0
120.1
0.9
0.9
2.9
1.1
-.4
.1
-.1
1.6
-.2
2.5
1972-92 .........
1974-92 .........
1967-72 .........
1972-77 .........
1977-82 .........
1982-87 .........
1987-92 .........
6.6
2.9
-.5
-.1
1.6
5.6
3.2
-2.2
2.4
2.7
-1.6
-2.2
-.5
-2.9
-.9
-2.3
-4.1
-1.1
SOURCE: Tables 14, 20, and 41
Enterprise Service—91
Chart 4. State and local government and private electric
power utility labor productivity indexes, 1967-92
Index, 1967 =100
200
200
180
180
Government
160
160
140
140
Private
120
120
100
100
80
1967
1972
1977
1982
1987
80
1992
Chart 5. Local government and private natural gas utility
labor productivity indexes, 1974-92
Index,1974 =100
105
105
100
100
95
95
90
90
85
85
Government
80
75
80
75
Private
70
70
65
65
60
1974
1977
1980
1983
1986
1989
60
1992
Chart 6. State government and private liquor store labor
productivity indexes, 1972-92
Index, 1972 =100
130
130
Government
120
120
110
110
100
100
Private
90
1972
Enterprise Service—92
1977
1982
1987
90
1992
Endnotes
1
Government Transactions, “Methodology Papers: U.S. National Income and Product
Accounts,” (BEA-MP-5), Washington: U.S Bureau of Economic Analysis, November, 1988,
p. 111.
2
A somewhat dated, but more expansive, discussion of the measurement of State and local
government electric power productivity is presented in BLS Bulletin 2166. See U.S. Bureau of
Labor Statistics, Measuring Productivity in State and Local Government, (U.S. Department of
Labor, Bureau of Labor Statistics), Bulletin 2166, December 1983, pp. 25-34.
3
A “municipal” power utility is defined by the U.S. Department of Energy as “a city, county,
irrigation district, drainage district, or other political subdivision or agency of a State competent
under the laws thereof to carry on the business of developing, transmitting, utilizing or distributing power.” See Financial Statistics of Major U.S. Publicly Owned Electric Utilities 1992, (U.S.
Department of Energy, Energy Information Administration, 1994) p. 532.
4
Financial Statistics of Major U.S. Publicly Owned Electric Utilities 1992, (U.S. Department of
Energy, Energy Information Administration, 1994) p. 512.
5
Public Power, January/February, 1994, p. 74.
6
The Bureau of Census does not list any government electric power employment for Montana in
1987, but the U.S. Department of Energy shows one small system operated by the City of Troy
in 1992. See Electric Sales and Revenue—1992, (Energy Information Administration, 1994), p.
57.
7
Computed from 1987 Census of Governments—Compendium of Government Finances (Bureau of the Census, 1989), pp. 88 and 93; and 1987 Census of Governments—Compendium of
Public Employment (U.S. Bureau of the Census, 1979), pp. 32 and 37.
8
Computed from Public Power, January/February, 1994, p. 72, and Financial Statistics of Major U.S. Publicly Owned Electric Utilities 1992, (U.S. Department of Energy, Energy Information Administration, 1994) pp. 4 and 504.
9
Standard Industrial Classification Manual (U.S. Office of Management and Budget, 1987), p.
284.
10
William Iulo, Electric Utilities—Costs and Performance, (Pullman, Washington: State University Press, 1961), p. 30.
11
“Gas and Electric Utilities—SIC 491, 492, 493,” unpublished technical note, Office of Productivity and Technology, U.S. Bureau of Labor Statistics, July, 1989, and Productivity Measures for Selected Industries and Government Services, Bulletin 2440, Washington: U.S. Government Printing Office, March 1994, p. 81.
12
William Iulo, Electric Utilities—Costs and Performance, (Pullman, Washington: State University Press, 1961), p. 30.
13
In 1995 BLS switched to the Tornqvist procedure to calculate its industry productivity measures as noted in Chapter 2. The data presented here reflect the computational procedures used
in 1992 and prior years. For further discussion see Kent Kunze, Mary Jablonski and Virginia
Klarquist, “BLS modernizes industry labor productivity program,” Monthly Labor Review, July
1995, pp. 3-12.
14
Financial Statistics of Major U.S. Publicly Owned Electric Utilities 1992, (U.S. Department
of Energy, Energy Information Administration, 1994) pp. 398 and 401.
15
With the deregulation of the electric power industry this situation is likely to change as generation, transmission and distribution are handled by different firms.
16
Personal communication from Diane Moody, American Public Power Association, July 1,
1994.
17
See Public Power, annual January/February issue.
18
See Electric Sales and Revenue—1992, (Energy Information Administration, 1994), p. 13.
19
U.S. Bureau of the Census. Public Employment: 1992, Series GE/92-1. U.S. Government
Printing Office, Washington, D.C., 1994, p. 5.
Enterprise Service—93
20
There is a question concerning the difference in the annual rate of change between 1982 and
1987 for total employment and full-time equivalent employment. The size of this difference is
apparently due to rounding and statistical discrepancies.
21
See table 224 for government statistics and table 149 for private and cooperative utility statistics, Productivity Measures for Selected Industries and Government Services, Bulletin 2440,
Washington: U.S. Bureau of Labor Statistics, 1994.
22
Ibid.
23
Gas Facts-1991 Data, Arlington, Virginia: American Gas Association, 1992, pp. 69-70 and
173.
24
Standard Industrial Classification Manual, (U.S. Office of Management and Budget, 1987),
p. 284.
25
Gas Facts—1991 Data, Arlington, Virginia: American Gas Association, 1992, p. 2.
26
Personal communication from Steve Owens, U.S. Bureau of the Census, September 9, 1994.
27
1987 Census of Governments—Compendium of Public Employment (U.S. Bureau of the Census, 1989), pp. 33-83, and 1987 Census of Governments—Compendium of Government Finances
(U.S. Bureau of the Census, 1990), pp. 66-74.
28
Purchased gas accounted for 50 percent of residential and 60 percent of commercial operating
costs for investor-owned utilities in 1991. Natural Gas 1992: Issues and Trends, Washington:
U.S. Energy Information Agency, March, 1993, pp. 74-6.
29
Payroll accounted for 15 percent of investor-owned utility operations in 1991. Gas Facts—
1991 Data, Arlington, Virginia: American Gas Association, 1992, p. 168.
30
The American Gas Association and others, classify all local government gas utilities as “municipal” utilities. In this bulletin, the term “municipal” is used in its more restricted sense, that is,
one of several types of local government.
31
A British thermal unit (BTU) is the quantity of heat that must be added to one pound of water
to raise its temperature one degree Fahrenheit under fixed temperature and pressure. A cubic
foot is 1,025.8 BTU’s.
32
“Gas and Electric Utilities—SIC 491, 492, 493,” unpublished technical note, Office of Productivity and Technology, U.S. Bureau of Labor Statistics, July, 1989, and Productivity Measures for Selected Industries and Government Services, Bulletin 2440, Washington: U.S. Government Printing Office, March 1994, p. 81.
33
BLS recently switched to the Tornqvist procedure to calculate its industry productivity measures as noted in chapter 2. The data presented in this bulletin reflect the computational procedures used in 1992 and prior years. For further discussion see Kent Kunze, Mary Jablonski and
Virginia Klarquist, “BLS Modernizes Industry Labor Productivity Program,” Monthly Labor
Review, July 1995, pp. 3-12.
34
U.S. Bureau of the Census. Public Employment: 1992, Series GE/92-1, Washington: U.S.
Government Printing Office, Washington, DC 1994, p. 6.
35
See table 150. Productivity Measures for Selected Industries and Government Services, Bulletin 2440, Washington: U.S. Government Printing Office, March 1994.
36
“Drinking Water: Stronger Efforts Essential for Small Communities to Comply with Standards,” Washington: U.S. General Accounting Office, March 1994, pp. 8-9.
37
“Drinking Water: Stronger Efforts Essential for Small Communities to Comply with Standards,” Washington: U.S. General Accounting Office, March 1994, p. 9.
38
“Environmental Investments: The Cost of a Clean Environment,” Report of the Administrator
of the Environmental Protection Agency to the Congress of the United States, EPA-230-11-90083, Washington: U.S. Environmental Protection Agency, November, 1990, p. F-16.
39
U.S. Bureau of the Census, Government Finances: 1991-92, Series GF92/5, Washington:
U.S. Government Printing Office, 1996, p. 39 and U.S. Bureau of the Census, Public Employment: 1992, Series GE/92-1, Washington: U.S. Government Printing Office, 1994, pp. 5-8.
Enterprise Service—94
40
These statistics exclude the District of Columbia. See U.S. Bureau of Census, 1992 Census of
Governments—Government Organization, Washington: U.S. Government Printing Office, 1994,
pp. 12-21.
41
William Fox, Size Economies in Local Government Services: A Review, Washington: U.S.
Department of Agriculture, 1980, pp. 26-30.
42
Robert M. Clark, The Safe Drinking Water Act: Its Implications for Planning,” in David Holz
and Scott Sebastian (eds.), Municipal Water Systems: A Challenge for Urban Resource Management, Bloomington: Indiana University Press, 1978, pp. 117-37.
43
H. Youn Kim and Robert M. Clark, “Input Substitution and Demand in the Water Supply
Production Process,” Water Resources Research, Vol 23. No. 2, pp. 239-44, February 1987.
44
U.S. Bureau of Census, Government Finances: 1991-92, Series GF92/5, Washington: U.S.
Government Printing Office, 1996, p. 16.
45
James I. Gillean and others, “The Cost of Water Supply and Water Utility Management,”
Contract Report 60-03-2071 (Cincinnati: U. S. Environmental Protection Agency, 1977), p. 9.
46
Robert M. Clark and James I. Gillean, “The Cost of Water Supply Utility Management,”
Environmental Modeling and Simulation, Proceedings of the Conference, Cincinnati, Ohio: U.S.
Environmental Protection Agency, 1976, pp. 808-813.
47
U.S. Office of Management and Budget, Standard Industrial Classification Manual, Washington: U.S. Government Printing Office, 1987, p. 285.
48
“Drinking Water Program: States Face Increased Difficulties in Meeting Basic Requirements,”
RCED-93-144, Washington: U.S. General Accounting Office, June 1993, p. 6.
49
Robert M. Clark, “Small Water Systems: Role of Technology,” Journal of the Environmental
Engineering Division, American Society of Civil Engineers, Vol. 106, No. EE1, Proceedings
Paper 15181, February 1980, p. 25.
50
Computed from community water system sample data as collected by Temple, Barker, and
Sloane for 1975 and 1980 and Research Triangle Institute for 1985. Research sponsored by
Office of Drinking Water, U.S. Environmental Protection Agency.
51
“Drinking Water: Compliance Problems Undermine EPA Programs as New Challenges Emerge,”
GAO/RCED 90-127, Washington: U.S. General Accounting Office, June, 1990, p. 53.
52
“Environmental Investments: The Cost of a Clean Environment,” EPA-230-11-90-083, Washington: U.S. Environmental Protection Agency, November, 1990, p. F-16.
53
“Environmental Investments: The Cost of a Clean Environment,” EPA-230-11-90-083, Washington: U.S. Environmental Protection Agency, November, 1990, pp. 4-16 to 4-24.
54
Harry P. Hatry, et. al., Efficiency Measurement for Local Government Services, Washington:
Urban Institute, 1979, pp. 35-42, and Harry P. Hatry, et. al., Service Efforts and Accomplishments Reporting: Its Time Has Come. Norwalk, Connecticut: U.S. Government Accounting
Standards Board, 1990.
55
Robert M. Clark, John A. Machisko and Richard G. Stevie, “Cost of Water Supply: Selected
Case Studies,” Journal of The Environment Engineering Division, February, 1979, p. 90.
56
John E. Schefter, “Domestic Water Use in the United States, 1960-85,” National Water Summary 1987 - Hydrologic Events and Water Supply and Use, Washington: U.S. Government Printing
Office, 1990, p. 75.
57
“Additional Federal Aid for Urban Water Distribution Systems Should Wait Until Needs are
Clearly Established,” Washington: U.S. General Accounting Office, November 24, 1980, p. 30.
58
John E. Schefter, “Domestic Water Use in the United States, 1960-85,” National Water Summary 1987—Hydrologic Events and Water Supply and Use, Washington: U.S. Government Printing Office, 1990, p. 79.
59
“Environmental Investments: The Cost of a Clean Environment,” EPA-230-11-90-083, Washington: U.S. Environmental Protection Agency, November, 1990, pp. F16-20.
Enterprise Service—95
60
Gary L. Rutledge and Christine R. Vogan, “Pollution Abatement and Control Expenditures,
1972-92,” Survey of Current Business, May, 1994, pp. 36-49.
61
Wayne B. Solley, Robert R. Pierce and Howard A. Perlman, Estimated Use of Water in the
United States in 1990, USGS Circular 1081, Washington: U.S. Government Printing Office,
1993.
62
U.S. Bureau of the Census, Public Employment: 1992, Series GE/92-1, Washington: U.S.
Government Printing Office, 1994, pp. 5-8.
63
U.S. Bureau of the Census, City Employment: 1992, Series GE/92-2. Washington: U.S. Government Printing Office, 1994.
64
1993 Transit Fact Book (American Public Transit Association, November, 1993), p. 33.
65
Ibid.
66
U.S. Bureau of the Census, Government Finances: 1991-92, Series GF/92-5. Washington:
U.S. Government Printing Office, Washington, D.C. 1996, pp. 10 and 16.
67
“National Transit Summaries and Trends,” (as taken from the 1992 National Transit Database), Washington: U.S. Federal Transit Administration, May, 1994, p. 3.
68
Ibid.
69
National Transit Summaries and Trends, (as taken from the 1992 National Transit Database),
Washington: U.S. Federal Transit Administration, May, 1994, pp. 15, 24-25 and 47.
70
National Transit Summaries and Trends, (as taken from the 1992 National Transit Database),
Washington: U.S. Federal Transit Administration, May, 1994, pp. 26-27.
71
National Transit Summaries and Trends, (as taken from the 1992 National Transit Database),
Washington: U.S. Federal Transit Administration, May, 1994, pp. 28-29.
72
National Transit Summaries and Trends, (as taken from the 1992 National Transit Database),
Washington: U.S. Federal Transit Administration, May, 1994, pp. 30-31.
73
National Transit Summaries and Trends, (as taken from the 1992 National Transit Database),
Washington: U.S. Federal Transit Administration, May, 1994, pp. 34-35.
74
National Transit Summaries and Trends, (as taken from the 1992 National Transit Database),
Washington: U.S. Federal Transit Administration, May, 1994, pp. 3, 15, and 32-33.
75
Standard Industrial Classification Manual, 1987, Washington: U.S. Office of Management
and Budget, 1987, p. 267.
76
Gordon J. Fielding, Roy E. Glauthier and Charles A. Love, Development of Performance
Indicators for Transit, (Irvine, California: University of California, Institute of Transportation
Studies, 1977); Wanda A. Wallace, “Mass Transit,” chapter 7, in Harry P. Hatry, et. al., Service
Efforts and Accomplishments Reporting: Its Time Has Come, Norwalk, Connecticut: Governmental Accounting Standards Board, 1990.
77
Gordon J. Fielding, Roy E. Glauthier and Charles A. Love, Development of Performance
Indicators for Transit, (Irvine, California: University of California, Institute of Transportation
Studies, 1977), p. 32.
78
B. M. Deakin and T. Seward, Productivity in Transportation (Cambridge, England: Cambridge University Press, 1969); John W. Kendrick, “Productivity Trends in U.S. Transportation
Industries,” as cited in Scheppach and Woehlcke; Raymond C. Scheppach, Jr. and L. Carl
Woehlcke, Transportation Productivity (Lexington, MA: Lexington Books, 1975); and U.S.
Bureau of Labor Statistics, Productivity Measures for Selected Industries and Government Services, Bulletin 2440, (Washington: U.S. Government Printing Office, March 1994).
79
These statistics are for total transit employment, private and government, as presented in the
1993 Transit Fact Book (American Public Transit Association, November, 1993), p. 98.
80
For further details see table 52, p. 71 of U.S. Bureau of Labor Statistics, Measuring Productivity in State and local Government, BLS Bulletin 2166, Washington: U.S. Government Printing
Office, December 1983.
Enterprise Service—96
81
The sample coverage reflects total sample employment as a percent of total local government
employment reported in the U.S. Bureau of the Census, Public Employment: 1992, Series GE/
92-1. U.S. Government Printing Office, Washington, DC. 1994, p. 6.
82
U.S. Bureau of Census, Public Employment: 1992, Series GE/92-1, U.S. Government Printing Office, Washington, DC 1994, pp. 5-8.
83
A dated, but more detailed discussion of the measurement of State alcoholic beverage control
sales is presented in BLS Bulletin 2166. See U.S. Bureau of Labor Statistics, Measuring Productivity in State and Local Government, (U.S. Department of Labor, Bureau of Labor Statistics), Bulletin 2166, December 1983, pp. 34-42.
84
North Carolina, a control State, is included in these statistics although all sales are by local
authorities.
85
U.S. Bureau of the Census. Public Employment: 1992, Series GE/92-1. U.S. Government
Printing Office, Washington, D.C. 1994, p. 24 with adjustments; and Government Finances:
1991-92, U.S. Government Printing Office, Washington, D.C. 1994. p. 20.
86
Standard Industrial Classification Manual(1987), Washington: U.S. Office of Management
and Budget, 1987, p. 329.
87
Standard Industrial Classification Manual (1987), Washington: U.S. Office of Management
and Budget, 1987, p. 310.
88
U.S. Bureau of the Census. Government Finances: 1990-91, Series GF/91-5, U.S. Government Printing Office, Washington, DC, 1993, p. A-5.
89
U.S. Bureau of the Census. Public Employment: 1992, Series GE/92-1. U.S. Government
Printing Office, Washington, D.C. 1994, p. A-4.
90
A discussion of the strengths and weaknesses of the individual measures is presented in Bulletin 2166. See U.S. Bureau of Labor Statistics, Measuring Productivity in State and Local
Government, (U.S. Department of Labor, Bureau of Labor Statistics) Bulletin 2166, December
1983, pp. 38-39.
91
The BLS recently switched to the Tornqvist procedure to calculate its industry productivity
measures as noted in Chapter 2. The statistics presented here reflect the computational procedures used in 1992 and prior years. For further discussion see Kent Kunze, Mary Jablonski and
Virginia Klarquist, “BLS Modernizes Industry Labor Productivity Program,” Monthly Labor
Review, July 1995, pp. 3-12.
92
BLS Handbook of Methods, U.S. Bureau of Labor Statistics, September 1992 (BLS Bulletin
2414) and James D. York, “Retail Liquor Stores Experience Flat Trend in Productivity,” Monthly
Labor Review, February, 1987, pp. 25-29.
Enterprise Service—97
Chapter 4. Corrections
T
his chapter discusses the measurement of labor productivity for State and local
government correctional services. Correctional services, in contrast to the enterprise services, are a general government service that is supported almost entirely by general revenue. There is no marketplace price, outputs are debatable, and
managers lack control over the demand for their services. There are even questions as
to what is included in correctional services.
The Bureau of the Census defines corrections as the “confinement and correction of
adults and minors convicted of offenses against the law, and pardon, probation, and
parole activities.”1 For purposes of BLS productivity statistics, corrections include all
the activities undertaken by State and local government employees assigned to the field.
All States, and many local governments, operate correctional programs. Corrections account for about 3.5 percent of all State and local government employment and
4.0 percent of all full-time-equivalent employees. It is one of the fastest, if not the
fastest, growing government service. In 1967, there were about 124,000 State and local
government employees in this service. By 1992, the number had reached 543,000, an
average annual growth of 6.1 percent.2
State and local governments spent over $28.7 billion in fiscal 1992 on corrections.
Capital expenditures accounted for about 14 percent of the total, operations for the
other 86 percent. More than three-quarters of the operational expenditures (77 percent)
went to labor compensation (salaries, wages, and benefits); the rest was devoted to the
purchase of supplies, contract services, and the like (table 44). Most productivity studies of corrections flow from an examination of cost functions. As noted in chapter 2,
the cost function is the dual of the production function. Since most correction managers
control their organization’s costs, but not their outputs or the type of programs provided, the cost function is the appropriate structure for measuring productivity. Also,
cost data are generally available or can be estimated for most correctional facilities.
Table 44. Expenditures of State and local government on corrections, fiscal
year 1992
(millions of dollars)
Expenditure category
Total
State
Local
Total ......................................
$28,701
$18,401
$10,300
Capital:
Total ...................................
Construction ......................
Other .................................
3,885
3,427
458
2,276
1,982
294
1,609
1,445
164
Current operations:
Total ...................................
Compensation ...................
Other .................................
24,815
19,146
5,669
16,125
12,349
3,776
8,691
6,797
1,894
NOTE: Totals may not add because of rounding.
SOURCE: U.S. Bureau of the Census, Government Finances: 1991-92 and Public Employment 1992; compensation and other estimated
Corrections—98
The number of adults under State or local government correctional supervision, either incarcerated or under supervised release (probation or parole) reached 4.6 million
in 1992.3 In addition, there were about 60,000 juveniles held in State and local government facilities, and an unknown number in aftercare or on probation. The number of
known offenders has increased dramatically since 1967, the first year of the BLS indexes. The exact numbers are not known, but rough calculations suggest that the increase in adult offenders is about 7 percent per year.
These data are for the entire correctional field. Resource and offender estimates for
the individual parts, such as prisons and jails, are presented later in this chapter.
Four basic parts of corrections services are reviewed here: State adult correctional
institutions, local jails, adult probation and parole, and juvenile correctional institutions. State adult correctional institutions are primarily State prisons, but include some
State-operated jails and community facilities such as halfway houses. This bulletin uses
the terms adult correctional institutions and prisons interchangeably.
On the following pages each of the four parts is briefly discussed, potential output
measures are examined, data to calculate indexes are presented, and indexes are calculated for prisons, jails, and juvenile correctional institutions. It is not possible to calculate an index for adult probation and parole services because of insufficient data. The
final section of this chapter presents findings and conclusions.
State Adult
Correctional
Institutions
State adult correctional institutions are the most important part of State and local
government correctional services, at least in so far as employment and cost are concerned. Also, it is one of the more rapidly growing State services. Between 1967 and
1992, the number of inmates held by the States increased by 325 percent. By the end of
1992, over three-quarters of a million inmates were held in State facilities.
Institutional setting
State governments spent almost $10 billion to operate their adult facilities in 1990,
the last year for which data are available. The annual average cost per prisoner in 1990
was about $15,600. However, the true cost was somewhat higher because of the cost of
uncounted State expenditures, such as benefits and services supplied by non-correctional State agencies, such as departments of health and education. In addition to operating costs, State governments spent another one-half billion dollars on capital facilities.4
Every State and the District of Columbia operates correctional institutions. For
most States, this means operation of State prisons, and in some cases, operation of
community facilities. Moreover, for seven jurisdictions—Alaska, Connecticut, Delaware, District of Columbia, Hawaii, Rhode Island and Vermont—this also means operation of jails, normally a local government function.
In 1990 there were about 246,000 State employees assigned to State adult correctional institutions. This is 75 percent of all State correctional employees. The remaining State correctional employees were assigned to probation, pardon, parole, juvenile
activities, overhead, and miscellaneous activities. State adult correctional institutions
accounted for about 73 percent of State correctional budgets in 1990. 5
The States with the largest number of inmates and the largest annual expenditures
are those with the largest population, that is California, Texas, and New York. California, alone, held over 90,000 inmates, operated 100 facilities, and spent over $2 billion
dollars on its prisons in 1990. At the other end of the scale, North Dakota operated two
facilities with fewer than 600 inmates, and spent less than $10 million.6
Labor is the most important resource, by far, in the operations of State adult institutions. One very detailed examination of New York State prisons concluded that employee costs accounted for about 80 percent of State operating costs in fiscal 1978.7
Another study, this one of the Virginia prison system, found that approximately 70
percent of prison operating costs in 1985 were allocated to personnel services and benefits.8 A study of selected Federal and State prisons found that personnel costs varied
Corrections—99
between 65 and 93 percent of total operating costs, depending on the facility, with the
average being 75 percent.9 In short, while overall comprehensive national statistics are
lacking, the readily available evidence suggests that labor comprises the largest part of
State institutional operating costs. Consequently, it should be a good measure of the
resources used to operate State adult correctional institutions.
The States operated, or were responsible for operating, 1,207 facilities in 1990, the
last year for which detailed statistics on prison operations are available. Most of the
facilities were general confinement prisons, but the States also operated reception and
diagnostic centers, farms, road camps, boot camps, and medical treatment centers. About
97 percent of all inmates are housed in these facilities. Another 3 percent are housed in
the 250 State operated or supervised community-based facilities that served as work
release and pre-release centers.
Large confinement facilities dominate the State prison scene with more than half of
all inmates confined in facilities having 1,000 or more prisoners.10 Larger facilities
generally require less labor per prisoner to operate than the smaller ones. Facilities
with 1-499 inmates employ 1 person for every 2.2 inmates; facilities with 500-999
inmates employ 1 person for every 2.5 inmates; and facilities with 1,000 or more inmates employ 1 person for every 3.1 inmates. These same general relationships hold
for individual occupational groups. That is, there are fewer custodial and security,
treatment and educational, maintenance and food service, and clerical and administrative personnel per inmate in the larger prisons than in the smaller institutions relative to
the prison population.11
State facilities are often categorized by level of security requirements—maximum,
medium, and minimum—and the inmates are evaluated and assigned accordingly. In
1990, 38 percent of the inmates were assigned to maximum security facilities, 49 percent to medium and 13 percent to minimum. Most (about 95 percent) inmates are
male.12
The courts play a major role in prison operations. Prior to the 1960s, the courts did
not insert themselves into this arena. In 1974, a U.S. Supreme Court decision changed
this when it upheld a lower court ruling that a prisoner is not “stripped of (his) constitutional protection when he is imprisoned for crime.”13 Over the past two decades, the
courts have issued a number of remedial orders, and as of June 1990, 323 State facilities were under some type of order or court imposed edict. Court actions focus on
crowding, medical and health practices, food service, sanitation conditions, and due
process.14 They have had a major impact on prison staffing and labor productivity.
The measurements covered in this discussion include most of the correctional institutions operated in the 50 States by State governments and the District of Columbia.
They are classified by the SIC as correctional institutions, SIC 9223, and defined as
follows:
Government establishments primarily engaged in the confinement and correction of offenders sentenced by a court. Private establishments primarily
engaged in the confinement and correction of offenders sentenced by a court
are classified in Services, Industry 8744. Halfway houses for ex-convicts
and homes for delinquents are classified in Services, Industry 8361. (The
following types of facilities are included in this SIC:)
Correctional institutions
Detention centers
Honor camps
Houses of correction
Jails15
Penitentiaries
Prison farms
Prisons
Reformatories
Because jails run by local governments are excluded from the accompanying indexes
for 44 States, only part of SIC 9223 is included in this measure.
Corrections—100
In addition to the adult correctional institutions, community institutions (e.g., halfway houses) operated by State governments are included. They are part of SIC 8361,
residential care:
Establishments primarily engaged in the provision of residential social and
personal care for children, the aged, and special categories of persons with
some limits on ability for self-care, but where medical care is not a major
element. Included are establishments providing 24-hour year-round care
for children. Boarding schools providing elementary and secondary education are classified in Industry 8211. Establishments primarily engaged in
providing nursing and health-related personal care are classified in Industry Group 805.16
The focus here is on State adult correctional institutions. State juvenile operations,
adult parole, pardon and probation operations, Federal facilities, and most local correctional operations are excluded. Six States and the District of Columbia operate local
jails, and these operations are included because the employees are included in the statistical counts.17 The following discussion sometimes uses the term “prisons” when
referring to all State adult correctional institutions.
Outputs
This section discusses the measures that might be used to calculate the output of
State prisons and community confinement facilities, reviews the data available to compute the measures, selects the preferred measure, and calculates the index.
Candidate measures
There are literally dozens of indicators that are routinely used to measure prison
operations and assess the performance of State adult correctional institutions.18 Three
of them are discussed here. But first, because they are so often discussed in the correctional literature, prison outcomes are briefly examined.
Outcomes assess the results of government activity. For prisons, outcomes might
measure the rate of recidivism. Specific measures include:
• Number or percent of former inmates who are incarcerated within 3 years of
release;
• Number of convictions within 3 years of release;
• Percent of time employed within 5 years of release;
• Average weekly earnings the first year following release;
• Percent who work in a job using the skills learned in prison.
While important, these are all measures of outcomes or the results of correctional work.
They are not the work of correctional institutions.
Adult correctional institutions process prisoners into facilities, test them, house and
feed them, guard them, educate and train them, and process them out of facilities. One
study characterized these activities as “hotel services.” While some analyses attempt to
measure individual activities such as these, most simply use the number of prisoners, or
some variation, as the measure of institutional output.19 Moreover, most governments
simply use a ‘set dollar cost per inmate per day’ to reimburse other governments or
private firms to house their prisoners. There is considerable prisoner contracting because of facility crowding, as discussed elsewhere in this chapter.20 In short, the number of prisoners is a reasonable proxy measure for the work performed.
The three output measures reviewed in this section are:
• Number of inmates incarcerated
• Number of inmates incarcerated differentiated by level of security
• Number of inmates incarcerated differentiated by program service.
The criteria used to evaluate each measure are presented in chapter 2.
Corrections—101
The first output measure to be considered is the number of inmates incarcerated. If
there is a single goal or objective of State correctional institution operations it is incarceration. Politicians, academics, the courts, and the general public all seem to agree on
this objective, and most institutional activities and processes, including housing, food
service, medicine, and recreation, go to support it. Even inmate rehabilitative activities, such as counseling, education, and training, are part of the incarceration process.
Most prison officials strongly support rehabilitative activities for their positive influence on inmate behavior while in confinement. That is, they view such programs as an
extension of their control mechanisms.
This measure, the number of inmates incarcerated, satisfies all the criteria listed in
chapter 2. It is measurable, repetitive, accurate, easily understood, and the final output
of the correctional institution. Also, there seems to be a good relationship between the
number of inmates and institutional staffing. That is, staffing is dictated by:
the characteristics and needs of (the) inmate population; the level of security required; the type of work, training, or rehabilitative programs provided; the physical layout of the facility; scheduled work hours, shift arrangements, and leave provisions; and (others).21
There are several variations of this basic measure. One uses an end-period count
(such as a year-end) as contrasted with a count of the average number of inmates held
over the period (the average daily population count). For conceptual reasons, the average daily count measure is preferred because it tracks fluctuations in the number of
prisoners accommodated. It should be more closely related to the resources expended
than the year-end measure. Irrespective of which measure is used, the same measure
should be used from year to year. Also, a comparable resource or labor measure should
be used. That is, a year-end inmate index should use a year-end labor measure and an
average daily inmate count should be matched with an average daily labor measure.
Another variation of the measure distinguishes between the number of prisoners
actually housed and the number under the jurisdiction of the State. In the latter case,
another jurisdiction or organization such as a local jail, a prison in another State, a
private correctional facility, or a Federal prison can house prisoners. Because the focus
is on the work performed by State employees, an output measure should count only
those prisoners held by the State.
The second measure considered is the number of inmates incarcerated differentiated
by level of security. This measure expands on the preceding one by separately identifying the number of inmates assigned to each level of physical security. Security level
means the degree of control exercised over the assigned inmates, that is, minimum,
medium, and maximum. The reasons for identifying the inmates by security level is
that resource requirements (costs and labor) vary by security level and the ratio of
inmates assigned to the various security levels changes through time.
Conceptually, this measure is preferred over the simple count of incarcerated prisoners. That is, by differentiating the index the unit labor requirements spent in the production of different services is more accurately reflected. This measure satisfies the other
evaluation criteria presented in chapter 2 equally well.
Security is paramount in correctional facility operations.22 Security and order are
maintained in prisons by controlling the movement of inmates, searching inmates and
facilities, monitoring inmate assignments and performance, intervening in disputes among
prisoners, staffing guard towers, and so forth. The 1990 Bureau of Justice Statistics
employment data show that two-thirds of all employees are classified as security personnel,23 a figure that has remained fairly constant through time. Five prison surveys
between 1962 and 1990 show that the custodial employees make up 65-67 percent of
all employment.24
Although security levels and definitions vary from State to State, the general concepts are universal. Three levels are used in this discussion—maximum, medium and
minimum—following the lead of the Bureau of Justice Statistics:
Corrections—102
Maximum security is for the most violent offenders. Security requirements are paramount. Facilities are closed to the outside world, and double fences are common. Inmate movement is limited within the facility, and inmate monitoring and control takes
precedent over all other activities.
Medium security is the most difficult of the three levels to define. Generally, the
inmates placed in medium security are thought worthy of rehabilitation and they are
provided with intensive program services. But perimeter security and inmate control
remain of paramount concern.
Minimum security is for the least violent offender. The facilities are relatively open,
and inmate controls are built as much on trust as on physical control. Road and forestry
camps and farms are examples of minimum security facilities.25 Community facilities,
such as halfway houses, normally contain only minimum security inmates.
As noted above, security classification is important for productivity measurement
because the cost varies by type of confinement. In 1984, the operating cost per inmate
in a maximum security prison was about $11,300 annually while the cost in a halfway
house was about $8,000. There were 3.9 inmates per correctional officer in maximum
security prisons, 4.0 in medium, 4.7 in minimum, and 6.1 in community facilities.26 In
1990, the cost per State and Federal inmate in maximum confinement was $16,507,
medium was $16,095 and minimum was $11,833. The cost per inmate in State operated community-based facilities was $9,709.27
The percent of inmates assigned to each security level has changed through time
(table 45). The greatest growth occurred in medium security facilities that housed almost 48 percent of all inmates in 1990; in 1974, 34 percent were housed in medium
security facilities.
Table 45. Percent of inmates assigned by facility security classification,
selected years.
Security classification
Confinement facilities:
Maximum .....................
Medium .......................
Minimum ......................
Community facilities ........
1974
1979
1984
1990
39.5
33.8
21.9
4.8
40.6
36.4
19.0
4.1
42.3
42.9
11.4
3.4
37.1
48.0
12.4
2.6
SOURCE: 1974: Computed from data presented in 1976 Sourcebook of Criminal Justice Statistics, Washington: Government Printing Office, 1976, p. 244. 1979: Computed
from data presented in 1981 Sourcebook of Criminal Justice Statistics, Washington: Government Printing Office, 1981, p. 141. 1984: Computed from data presented in 1984 Census of State Adult Correctional Facilities, Washington: U.S. Department of Justice, 1986, p.
6. 1990: Computed from data presented in Census of State and Federal Correctional Facilities, 1990, Washington: Government Printing Office, pp. 9 and 20.
The custody level of the inmate usually, but not always, matches the physical security level of the facility to which the inmate is assigned. That is, maximum security
prisoners are generally held in maximum security facilities. But on occasion, maximum
security inmates can be held in a medium or even a minimum security facility. It is
much more common for minimum security prisoners to be held in medium or maximum
security prisons. Calculations require that inmates be differentiated by the physical
security level rather than by custodial security level because data are not available with
which to calculate weights by custodial security level.28
For productivity calculations, correctional institution output should differentiate inmates by the level of physical security provided and the output weighted by the appropriate base year unit labor requirement weights. To calculate facility output by security
level, the preferred labor measure is the number of custodians or guards. However,
because all employees are guards, to a certain extent, and the ratio of custodians to total
employment seems to be fairly constant through time, it should be acceptable to use
total employment when custodial data are unavailable.
Corrections—103
The final measure of the three, inmates incarcerated with differentiated program
service, incorporates services such as education, vocational training, counseling, recreation, library, and religion, into the basic measures discussed above. The reason for
including program services in the basic measure is that about 15 percent of State correctional institution resources go to support program services, and the labor used and services provided may vary through time. This measure should provide additional insights
into productivity movements because it further differentiates outputs.
Although data are lacking on the number of employees working in program services,
the 1990 Bureau of Justice Statistics prison survey shows that professional, technical,
and educational personnel accounted for 16.0 percent of U.S. State correctional facility
staff. In 1984, it was 15.3 percent and in 1979, 15.1 percent.29 A study of New York
corrections calculated that program services consumed about 17 percent of that State’s
prison budget in fiscal 1978 (prison industry and recreation services were added to
regular program services to reach this figure).30
Most, if not all, correctional institutions provide some program services. While
separately identifying each service and including it in the output index would be useful,
there are two problems with such an approach. First, no approach has yet been developed to measure outputs for many program services such as education and recreation.
Second, the necessary data are lacking to make the measurements, even in those cases
where there is a reasonable idea of what to measure.
Where there are proxy or surrogate measures and measurement data, the available
statistics suggest that program service outputs and inputs have increased at about the
same rate. Thus, there is no real reason to include, and several reasons to exclude,
program services from overall measurements. Furthermore, it is unlikely that any measurements would substantially affect an index of State correctional output, even if they
could be included at this time.
Crowding is a special concern and a perennial problem in correctional institutions.
It was discussed in the first annual report on State prisons in 1926, which noted that
correctional facilities were operating at 108 percent of capacity.31 The 1990 Bureau of
Justice Statistics survey notes that facility crowding ranged from 101 percent to 122
percent, depending on the capacity measure used. In June 1990, 172 of the 1,207 State
correctional facilities were operating under a court order or consent decree to limit the
number of inmates because of crowding.32
Crowding impacts many facets of prison operations. The concern here is how productivity is affected when more and more inmates are crowded into a facility? Some
argue staffing is not increased proportionally.33 That is, labor productivity would increase as inmates are crowded into a facility. Most research suggests otherwise. Generally, staff increases seem to parallel inmate growth. To quote one report:
There is some evidence that correctional systems may respond to pressures
of population growth by increasing the level of supervision over inmates.
Total staff increases nationally in State prisons between 1979 and 1984
were identical to the increase in the number of inmates (45 percent); however, since most of the personnel increase over the period was among correctional officers, the number of inmates per officer actually dropped from
4.6 to 4.1.34
Apparently, the individual facility unit labor cost curve decreases as long as the
number of inmates is less than its design capacity and remains constant once the design
capacity is reached. Because most facilities operate at and beyond capacity, a flat unit
labor cost curve becomes evident. Institutional considerations such as budget constraints, court rulings, press reports, and security concerns apparently serve to put an
upper bound on facility crowding.
Furthermore, some studies conclude that facility crowding has a generally negative
affect on institutional suicides and assaults. Other studies conclude that suicides and
assaults do not vary with increasing population but because of more important factors.
Corrections—104
To quote:
When density levels are compared with equivalent security grades, no clear
pattern emerges. The highest density maximum security facilities, for example, evidenced the highest rate of suicide but had a rate of homicide
lower than that reported in moderate density prisons and about the same as
that in low density prisons. Moreover, for prisons of each security level,
inmate-on-inmate assaults were most prevalent in the lowest density prisons. Similarly, institutional disturbances in minimum and medium security
facilities were most prevalent in prisons with the lowest population densities. In general, no consistent pattern emerges from these data indicating
that the incidence or prevalence of these negative events increases with
greater population densities.35
In short, the evidence and research on prison crowding is mixed, and suggests no
reason for explicitly considering prison crowding in prison output and productivity
indexes.
Inmates incarcerated: An
undifferentiated output
series
Inmates incarcerated: A
security differentiated
output series
Summary of output measures. The preferred conceptual measure of adult correctional
facility output is the number of inmates weighted by the type of prisoner and the type of
treatment provided. Unfortunately, data are lacking to develop even the simplest treatment modalities. This effectively restricts the calculations presented here to two output
measures. The first measure is a count of the number of inmates in State correctional
institutions. The second is a count of the number of inmates differentiated by the level
of facility security. The next two sections of this paper present calculations for these
two measures.
The undifferentiated output measure is the number of inmates held. Although this
measure is straightforward, there are several conceptual and data problems. As discussed earlier, there is the issue of whether to use an average daily or a year-end count.
The average daily count is generally preferred for conceptual reasons, but data are
lacking with which to compute such an index. Thus, the year-end count is used in the
measure.
Another issue is whether to count the number of persons under the jurisdiction of the
State or the number held in State facilities. As discussed earlier, for productivity purposes, the measure should count all inmates housed, irrespective of the circumstances
under which they are held. While the concept is simple, the problem, once again, is data
availability. Custodial data are not readily available prior to 1977. Prior to 1977, the
prisoner data are for felons under State jurisdiction.36
Examination of the custodial and jurisdictional series shows that they increase at
about the same rate, at least for the years for which comparable data are available. The
change from 1977-86 showed a similar but not identical movement. Thus, for trend
measurement purposes it does not make much difference whether a jurisdictional or
custodial index is used; level estimates present a different situation. It is important that
the same data series is used throughout the measured period or, if not, that the data
series are linked when shifts are made between data series. 37
Table 46 presents the undifferentiated output series and index. Given the issues just
discussed, the index is constructed using a year-end jurisdictional count for 1973-77
linked to the custodial index for 1977 forward.
A better measure of prison output for productivity measurement purposes is the number of inmates differentiated by facility security level (maximum, medium, and minimum). In addition, community facilities are separately identified because they have
very different unit costs. Community facilities are those where more than half of the
residents regularly leave, unaccompanied, for work, study, school, or other activities.
Community facilities are often described as half-way houses and generally house only
minimum security inmates.
Corrections—105
Table 46. State adult institution inmate indexes, 1973-92
Year
1973 ...................
1974 ...................
1975 ...................
1976 ...................
1977 ...................
1978 ...................
1979 ...................
1980 ...................
1981 ...................
1982 ...................
1983 ...................
1984 ...................
1985 ...................
1986 ...................
1987 ...................
1988 ...................
1989 ...................
1990 ...................
1991 ...................
1992 ...................
Felons,
year
end
Felon
index
(1973=100)
181,396
196,105
216,462
235,853
247,507
100.0
108.1
119.3
130.0
136.4
Custody
year
end
258,643
269,765
281,233
295,819
333,251
375,603
394,953
417,682
451,812
486,655
520,336
562,605
629,995
684,406
728,605
779,134
Custody
index
(1977=100)
100.0
104.3
108.7
114.4
128.8
145.2
152.7
161.5
174.7
188.2
201.2
217.5
243.6
264.6
281.7
301.2
Average annual
rate of change:
1973-77 ..............
1977-92 ..............
1973-92 ..............
Linked
inmate
index
(1973=100)
100.0
108.1
119.3
130.0
136.4
142.3
148.4
156.1
175.8
198.1
208.4
220.3
238.4
256.7
274.5
296.8
332.4
361.1
384.4
411.0
8.1
7.6
7.7
SOURCE: Historical Statistics on Prisoners in State and Federal Institutions, 1925-86 , table 1,
p. 12, and Correctional Population in the United States, table 5.18, selected years
The security-differentiated indexes were constructed using the inmate series discussed in the preceding section and inmate security data collected in the Bureau of
Justice Statistics surveys conducted for 1973, 1979, 1984, and 1990. Inmate security
data are not available prior to 1973. Separate indexes were constructed for each security level (table 47).
Annual growth rates of the inmate population produced by the calculations for the
various security levels are shown in the tabulation.
Maximum security
Medium security
Minimum security
Community security
7.4
9.7
4.5
4.3
All statistics are compounded and cover the 1973-92 period (table 47).
A security-differentiated index was calculated using these data. Separate segments
were calculated for 1973-79, 1979-84, 1984-90, and 1990-92 using the unit labor requirements, and the individual segments were combined using labor weights to create a
single index.
The unit labor weights used to combine the segments are presented in table 48. In
each instance they reflect the number of inmates divided by the number of custodial
personnel (e.g., guards). As table 48 shows, weights increase through time for each
security level except for minimum security in 1979. The reason for the drop in 1979 is
not known; it is possible that the problem is 1973, not 1979.
Corrections—106
Table 47. Unweighted output indexes by security level, 1973-92
(1973=100)
Year
Comparison of the output
indexes
Maximum
Medium
Minimum
Community
1973 ...................
1974 ...................
1975 ...................
1976 ...................
1977 ...................
1978 ...................
1979 ...................
100.0
108.6
120.5
131.9
139.1
145.8
152.7
100.0
109.4
122.3
134.8
143.2
151.1
159.3
100.0
105.7
114.0
121.2
124.2
126.3
128.3
100.0
105.5
113.5
120.5
123.1
125.0
126.7
1980 ...................
1981 ...................
1982 ...................
1983 ...................
1984 ...................
1985 ...................
1986 ...................
1987 ...................
1988 ...................
1989 ...................
162.0
184.0
209.1
221.6
236.3
250.3
263.9
276.0
291.8
319.4
173.6
202.4
235.7
255.9
279.1
307.9
338.1
368.4
405.7
462.7
124.2
127.9
130.5
122.9
114.9
126.0
137.6
149.1
163.4
185.4
128.7
139.8
151.8
153.5
155.9
162.2
167.7
171.8
177.7
189.9
1990 ...................
1991 ...................
1992 ...................
339.0
360.9
385.9
511.7
544.7
582.5
204.0
217.2
232.3
196.5
209.2
223.7
Average annual
rate of change:
1973-92 ..............
7.4
9.7
4.5
4.3
The preferred conceptual output measure is the number of inmates incarcerated differentiated by security level and program service. Unfortunately, data are lacking (and
in some instances the necessary theory) to differentiate program services.
Thus, the indicator used to measure adult correctional institution outputs is the number of inmates incarcerated differentiated by security level. Because of data limitations, it is impossible to differentiate the index prior to 1973, thus the index covers only
1973-92.
The average annual increase in output between 1973 and 1992 was 7.8 percent, and
there were increases every year (table 49). As noted earlier, the average annual increase by security classification varied from 9.7 to 4.3 percent (table 47). The unweighted
index, which is also shown in table 49, increased at almost the same average annual rate
(7.7 percent) as the weighted index.
Table 48. Unit labor requirement weights by facility-security level for
State government adult institutions, selected years
Security level
1973
1979
1984
1990
Confinement facility:
Maximum .............
Medium ...............
Minimum ..............
Community facility ....
.213
.209
.191
.158
.216
.232
.190
.161
.252
.240
.208
.163
.267
.243
.210
.197
SOURCE: 1973: Calculated from data taken from a special computer run of BJS
census data collected in 1974. 1979: Calculated from data taken from a special
computer run of BJS census data collected in 1979. 1984: Calculated from data
taken from 1984 Census of Adult Correctional Facilities, Washington: U.S. Bureau of
Justice Statistics, 1986, p. 7. 1990: Calculated from data taken from Census of
State and Federal Correctional Facilities, 1990, Washington: U.S. Government Printing Office, May, 1992, p. 20 and data provided by Jim Stephen, BJS, September 9,
1992.
Corrections—107
Table 49. Comparison of two State adult correctional output indexes,
1973-92
(1973 = 100)
Unweighted
Security
weighted
1973 .............................
1974 .............................
1975 .............................
1976 .............................
1977 .............................
1978 .............................
1979 .............................
100.0
108.1
119.3
130.0
136.4
142.3
148.4
100.0
108.2
119.5
130.3
136.9
142.9
149.0
1980 .............................
1981 .............................
1982 .............................
1983 .............................
1984 .............................
1985 .............................
1986 .............................
1987 .............................
1988 .............................
1989 .............................
156.1
175.8
198.1
208.4
220.3
238.4
256.7
274.5
296.8
332.4
157.3
177.7
200.9
212.0
224.9
243.2
261.9
279.9
302.6
338.7
1990 .............................
1991 .............................
1992 .............................
361.1
384.4
411.0
367.9
391.6
418.8
Average annual
rate of change:
1973-92 ........................
7.7
7.8
Year
Labor inputs
There were about 250,000 adult institutional employees in State governments in
1990, up from less than 50,000 in 1965. The number of employees is a relatively good
indicator of the resources expended on State prison operations because about 75 percent of all prison operating expenditures go to compensate prison employees. Furthermore, an index of prison employment is a good indicator of how the use of prison
resources changes through time.
Custodial employees or guards account for about two-thirds of the prison labor force,
a statistic that has remained relatively constant through time (table 50). In 1984, educational and treatment personnel including teachers, social workers, doctors, dentists,
nurses, psychiatrists, and psychologists made up another 16 percent. Maintenance and
food service workers accounted for almost 8 percent, clerical workers about 7 percent,
and administrative personnel about 4 percent.
Prison employment is 80 percent male as are the inmates, but the male-female ratio
varies by occupation or function. The guard or custodial occupation is 88 percent male,
the administrative staff is 82 percent male and the treatment/educational staff is 67
percent male. The clerical staff, on the other hand, is only 16 percent male.38
Most State adult prison employees are full-time employees. The criminal justice
dimensions of the job, such as the requirement for staff security clearances, for around
the clock operations, and for specialized job training, dictate that full-time employees
be used in most instances. Almost 98 percent of all State prison employees work full
time. The figure for guards, maintenance, and food service personnel is 99 percent; for
treatment and education it is 92 percent; and for medical it is 85 percent. By comparison, the figure for all State government employees is 75 percent.39
Two labor indexes, full-time-equivalent and total employment, are normally calculated for government productivity indexes. The two indexes often move in concert, and
this should be the case for prisons because most employees are full-time employees.
The labor index reflects the number of full-time-equivalent employees.
Corrections—108
Table 50. State government correctional institution employment by occupation,
1962, 1965, 1974, 1979 and 1984
Occupation
1965
1974
1979
1984
Total .............................. 42,721
46,680
60,604
95,724
141,958
Custodial ....................... 27,715
Guards .....................
26,966
1
749
Wardens ..................
30,809
30,809
39,298
38,157
1,141
59,383
59,383
94,601
94,601
6,319
2,851
1,341
365
181
614
967
14,492
9,444
21,678
13,405
482
169
933
1,289
Treatment and
education ...................
Teachers ...................
Social workers ..........
Psychologists ...........
Psychiatrists .............
Doctors .....................
Nurses ......................
Other medical ...........
Combined medical ...
1962
3,061
1,457
525
158
96
517
308
Other ............................. 11,945
3,1372
1,654
1,124
3063
327
327
3,464
6,993
12,734
14,987
21,849
25,670
1
Wardens not identified separately each year.
Data in subcategories do not add to total.
3
Includes physicians, psychiatrists, and dentists.
2
SOURCE: 1962—National Manpower Survey of the Criminal Justice System, Volume 3, Corrections, Washington: U.S. Law Enforcement Administration Agency, September, 1978, p. 15; as
taken from U.S. Bureau of Prisons Statistics. 1965—U.S. President’s Commission of Law Enforcement and Administration of Justice, Task Force Report: Corrections, Washington: Government Printing Office, 1967, p. 180. 1974—National Manpower Survey of the Criminal Justice
System, Volume 3, Corrections, Washington: U.S. Law Enforcement Administration Agency, September, 1978, p. 15; as taken from 1974 special survey of State Adult Prisons. 1979—U.S. Bureau
of Justice Statistics, “Census of State Adult Corrections Facilities, 1979,” Ann Arbor, Michigan:
Inter-University Consortium of Political and Social Research. Unpublished data taken from survey
documentation. 1984—1984 Census of State Adult Correction Facilities, NCJ-105585, Washington: U.S. Bureau of Justice Statistics, August 1987, and unpublished data taken from survey documentation.
There is no annual data series on State adult institutional employment. The procedure used to estimate State prison employment is to benchmark the year-to-year change
in State correctional employment as shown in the annual Census of Government. The
benchmarks are taken from the Presidential Commission on Law Enforcement statistics
for 1965 and the Bureau of Justice Statistics survey data for 1979, 1984, and 1990.
Comparison of Census and benchmark data show that for 1965-1979 the two measures
increased at about the same rate (less than 2 percent difference over the 14 years). Over
the 1979-84 period, the Bureau of Justice Statistics measurements increased about 9
percent more than the Census data as they did over the 1984-90 period. The benchmark
data include personnel paid by other State government agencies, such as departments of
education and health, but who work in prisons. The Census data probably exclude
these employees. These employees are less than 3 percent of total prison employment,
a figure that has been consistent over the years. Thus, it should not affect labor trends.
The average annual increase in State correction employment as calculated by the
Bureau of Census, and State adult institutional employment as estimated for this study,
are shown in table 51. The average annual increase in the Census data between 1967
and 1992 is 6.2 percent. The estimated increase in State adult institutions is 7.3 percent. For 1973-92, the period covered by the output and productivity indexes, the
average annual increase is 7.8 percent.
Corrections—109
Table 51. Comparison of State correctional employment and
State adult correctional institutional employment indexes,
1967-92.
(1967 = 100)
Year
Corrections
employment
Adult
institutional
employment
1967 .........................................
1968 .........................................
1969 .........................................
100.0
104.2
109.9
100.0
104.1
109.6
1970 .........................................
1971 .........................................
1972 .........................................
1973 .........................................
1974 .........................................
1975 .........................................
1976 .........................................
1977 .........................................
1978 .........................................
1979 .........................................
118.0
126.3
134.1
140.5
150.0
154.5
166.5
176.9
182.8
196.7
117.5
125.7
133.3
139.5
148.7
153.0
164.7
174.8
180.4
193.9
1980 .........................................
1981 .........................................
1982 .........................................
1983 .........................................
1984 .........................................
1985 .........................................
1986 .........................................
1987 .........................................
1988 .........................................
1989 .........................................
198.0
219.7
239.9
254.5
280.4
300.4
317.8
343.3
371.4
398.5
198.9
224.7
249.8
269.7
302.2
328.8
353.1
387.0
424.8
462.4
1990 .........................................
1991 .........................................
1992 .........................................
428.5
441.3
453.5
504.1
538.4
577.1
Average annual
rate of change
1967-92 ....................................
1973-92 ....................................
6.2
6.4
7.3
7.8
NOTE: Adult institutional employment calculated using annual Census data
and Bureau of Justice Statistics benchmarks. See text for discussion of procedures.
SOURCE: Corrections employment calculated from annual issues of Public
Employment (Bureau of the Census)
Productivity indexes
State adult correctional labor productivity as measured by output per employee, was
essentially flat between 1973 and 1992. Between 1973 and 1982 it increased 1.3 percent on an annual average basis as output growth outstripped labor. Over the next
decade, 1982-92, it decreased by 1.0 percent on an annual average basis. Both output
and labor input increased rapidly, although at different rates, throughout the measured
period (table 52).
Corrections—110
Table 52. State government adult correctional institution productivity indexes,
1973-92
(1970 = 100)
Output
index
Labor
index
Productivity
index
1973 .............................
1974 .............................
1975 .............................
1976 .............................
1977 .............................
1978 .............................
1979 .............................
100.0
108.2
119.5
130.3
136.9
142.9
149.0
100.0
106.6
109.7
118.1
125.3
129.3
139.0
100.0
101.5
109.0
110.4
109.2
110.5
107.2
1980 .............................
1981 .............................
1982 .............................
1983 .............................
1984 .............................
1985 .............................
1986 .............................
1987 .............................
1988 .............................
1989 .............................
157.3
177.7
200.9
212.0
224.9
243.2
261.9
279.9
302.6
338.7
142.6
161.1
179.1
193.3
216.7
235.7
253.1
277.4
304.5
331.4
110.3
110.3
112.2
109.6
103.8
103.2
103.5
100.9
99.4
102.2
1990 .............................
1991 .............................
1992 .............................
367.9
391.6
418.8
361.3
385.9
413.7
101.8
101.5
101.2
Average annual
rate of change:
1973-92 ........................
1973-82 ........................
1982-92 ........................
7.8
8.1
7.6
7.8
6.7
8.7
0.1
1.3
-1.0
1973-79 ........................
1979-84 ........................
1984-90 ........................
1990-92 ........................
6.9
8.6
8.6
6.7
5.6
9.3
8.9
7.0
1.2
-.6
-.3
-.3
Year
Local Jails
Jails and prisons are often treated and discussed as if they were the same, and in
many ways they are similar. Both hold prisoners, both have grown dramatically over
the past three decades, both report on their operations, and both share the same standard
industrial classification system number.
But, in many respects, the two institutions are very different. Jails hold convicted
persons and those awaiting trial while prisons hold only convicted persons. Jail inmates serve shorter sentences (usually less than 1 year) than prison inmates. Jails are
usually located in cities while prisons tend to be located in rural areas. Jails tend to be
smaller than prisons and offer far fewer treatment and education programs.40
Institutional setting
Local governments spent almost $6.9 billion to operate their jails in 1993; the cost
per prisoner per year was about $14,700. In addition to operating costs, local governments spent another $2.8 billion on capital costs to add to and modernize their jails the
same year.41
Many more individuals are sent to jail each year than to State prisons although the
number of persons confined to prison at any point in time is much larger than those in
jail. Jails are used for short-term detention and confinement and as a result have a high
turnover of persons compared to prisons. About half of those in jail are awaiting trial
while the rest are serving sentences or are being held for other jurisdictions.
The number of inmates in jails has increased over the years, as the number of jails
has decreased.
Corrections—111
Year
1970
1978
1983
1988
1993
Average number
of inmates
Number of
jails
40
45
67
104
139
4,037
3,493
3,338
3,316
3,304
The decrease in the number of jails is largely the result of consolidations and a
search for greater efficiency in operations. Yet, even in 1993, over half of the Nation’s
jails had 50 or fewer inmates.42
There is some question as to what constitutes a jail. As the term is used here, a jail is
a facility that holds inmates beyond arraignment, usually for more than 48 hours, and is
operated by local government employees. Jails are sometimes known as detention centers, county prisons, workhouses, or houses of correction. Specifically excluded from
the definition are temporary lockups that house individuals for less than 48 hours, Federal and State operated facilities, and privately operated facilities. The definition used
here is the one used by BJS in its data collection instruments.
Local governments operate most, but not all, jails. Alaska, Connecticut, Delaware,
Hawaii, Rhode Island, Vermont, and the District of Columbia operate unified systems,
that is, combined jail and prison systems. As noted in the discussion of State prisons,
statistics on these facilities are included in adult institutional tabulations.43
Also, some jurisdictions have turned to private companies to operate their jails. The
private sector has long been a source of supplies and services for jails, but it is only
recently that it has taken over the operation of entire facilities. The difficulty in constructing new jails, rising incarceration costs, and the use of “privatization” in other
public services, has generated interest in privatizing jail operations. And while there is
considerable discussion of the concept, only 17 jails were privately operated under
contract to local governments in 1993.44
Sheriffs operate the majority of the jails in the United States, 85 percent in the early
1980s according to one count. However, correctional administrators or wardens operate jails in most large cities and some counties in Pennsylvania. Also, elected jailers
operate jails in Kentucky, and jail administrators are appointed in New Jersey.45
Labor is a good measure of the resources used in jail operations because it accounts
for the major portion of jail operating expenditures, by one account 78 percent.46 A
study in Washington State found labor costs to be 76 percent of total jail expenditures,
supplies 15 percent, purchased services 8 percent, and capital 1 percent.47 The same
study found that the percent of expenditures devoted to personnel ranged from a high of
94 percent in one jurisdiction to a low of 40 percent in another.
The productivity measurements covered in this section include jail operations in
most States as specified by SIC 9223, correctional institutions:
Government establishments primarily engaged in the confinement and correction of offenders sentenced by a court. Private establishments primarily
engaged in the confinement and correction of offenders sentenced by a court
are classified in Services, industry 8744. Half-way houses for ex-convicts
and homes for delinquents are classified in Services, industry 8361.
Correctional institutions
Detention centers
Honor camps
Houses of correction
Jails48
Penitentiaries
Prison farms
Prisons
Reformatories
State prisons, penitentiaries, reformatories, and jail operations of six States and the
District of Columbia that were included in the preceding section are not covered here.
Corrections—112
Also, juvenile facilities and probation and parole activities are excluded from this section. Hence, only part of SIC 9223 is included in the indexes presented here.
Precise statistics are lacking as to what part of SIC 9223 resources go to support
State prison operations and what part goes to local jails. An estimate for 1990 suggests,
however, that operating costs were about 67 percent for State prisons and 33 percent for
local jails. The employment ratios are about the same, 65 percent in State prisons and
35 in percent local jails.49
One observer noted in 1975 that accurate national data on jail operations were not
readily available because most jails were run by local authorities, kept inadequate records
on their operations, and most had no requirement to report to a central authority.50
Since this observation was originally made, largely as a result of Bureau of Justice
Statistics efforts, the data situation has improved markedly. But there are still major
gaps in the data and very little research on jail operations, at least when compared with
prisons. Furthermore, data consistency is a real problem in formulating a long-term
local government jail productivity index.
Outputs
A search of the literature found little analysis or even discussion of how to measure
jail services or outputs.51 This section reviews some of the measures that have been
used for jail services, identifies the preferred measures, and calculates an output index.
Candidate measures
Six measures which have been used or suggested for use are considered here: Inmates confined, inmates confined by detention status, average daily population (ADP),
admissions and releases, number of confinement units or beds, and program services.
Most are surrogate measures for the work performed. The following section briefly
discusses each in terms of their strengths and weaknesses, using the criteria presented in
chapter 2.
Inmates confined. Probably the most frequently cited and used measure of jail output is
a count of the number of inmates held. Here it is used as a proxy for the work performed, that is, the greater the number of inmates the more work performed. The measure is sometimes broken down by sex, age (juvenile-adult), and inmate status (pretrial, convicted and/or awaiting sentencing, and serving sentence). This measure meets
most of the output evaluation criteria listed in chapter 2, that is, it is measurable, repetitive, and understandable. However, no statistics were found that showed the relationship between output and the resource units spent in their production although simple
calculations show a good correlation between the number of inmates and the number of
jail employees.
Although there are problems in using the inmate count as a measure of jail output,
the actual counts should be extremely accurate since physical counts of the number of
inmates are usually taken several times each day.
Inmates confined by detention status. A variation of the inmate count is a count differentiated by type of detention. For State prisons, inmates were categorized by security
level, that is, maximum, medium, minimum, and community detention, because research
shows that resource requirements vary by level. No comparable research was found in
the case of jails.
There are several different ways to categorize jails and jail inmates. One is by the
type of security such as that used in State prisons. Another is by type of inmate status:
Awaiting arraignment, convicted and awaiting sentencing, serving sentence, and technical probation and/or parole violation. Apparently, no one has yet determined how to
differentiate inmate counts by type of prisoner for productivity calculations and, until
that is done, further research on this measure cannot proceed.
Average daily population (ADP). This measure is commonly calculated and widely
used. It is a better measure than the 1-day inmate count when assessing output over an
extended period, such as a month or year, because it captures the change through time.
Corrections—113
The measure meets most of the evaluation criteria noted above. A 1976 statistical study
of 60 jails in California found that the ADP accounted for about 80 percent of the total
variation in annual operating cost.52
While ADP is commonly used, there, apparently, is no commonly agreed upon calculation procedure. Is the count taken every day and averaged? Is the count taken at 6
p.m., 12 midnight or 6 a.m. and considered the “daily” population for that day? Or is
the ADP a simple average of the beginning and ending period such as 1 year? There are
a host of possible definitions and measurement technicalities. Early surveys did not
specify the procedure for calculating the ADP but left it up to the reporting jurisdiction.
More recent Bureau of Justice Statistics surveys have asked reporting jurisdictions to
sum the daily counts and divide by 365.
Admissions and releases. The number of jail admissions and releases greatly exceeds
the number of prisoners which is the opposite of the State prison situation where the
number of prisoners held exceeds the number of admissions. Bureau of Justice Statistics’ jail data for 1993 show a ratio of 1:28 (number of admissions/average daily population), or 28 admissions for each inmate.53
The concern about admissions and releases is that they generate additional work that
affects productivity. The work of admissions and releases includes documentation of
prisoner history, prisoner interview, physical examination of the inmate, storage of personal effects, issuance and take back of jail clothing, movement of the prisoner, and so
forth. The work and time that is associated with each admission and release may be
small in absolute terms but for all inmates it may be important. More importantly, the
time spent on each prisoner may change through time, that is, the unit resource requirement may change. Also, the ratio between the number of admissions and the number of
inmates may change through time. Each of these considerations can affect productivity
trend calculations.
Examination of the readily available data suggests that admissions decreased as a
percent of the total number of inmates between 1983 and 1993. However, there is not
a great deal of difference between the data for 1988 and those for 1993. In 1983 there
were 36 admissions for each ADP inmate, in 1988 there were 29 and in 1993, as noted
above, there were 28.54
Given the large number of admissions to and releases from the Nation’s jails each
year and the apparent decrease in the number of admissions and releases compared with
the number of inmates, the work associated with admissions and releases probably should
be identified and counted separately. However, no studies were found that would suggest, in quantitative terms, the importance of this activity in terms of the resources
expended. Nor is it evident how one would calculate resource weights for this activity,
nor was any proxy measure found that could be used in lieu of resource weights. The
Bureau of Justice Statistics collects data periodically on the number of admissions and
releases in the Nation’s jails, but without estimates of the resources devoted to this
activity it is impossible to calculate an output index for it.
Number of confinement units and/or number of beds. Two other measures that are used
in measuring jail services, and for which data are often collected, are the number of
beds and the number of confinement units. The concept of a bed is simple, but a confinement unit can be anything from a single cell that can hold one person to a drunktank which can contain several hundred inmates. At one extreme a jail could have 100
beds and 1 inmate, at the other it could have 1 bed and 100 inmates with 99 inmates
sleeping on the floor.
Both measures are important in calculating crowding and planning for future facilities, and the courts have expressed an interest in both statistics. But they are measures
of capacity not output nor the work of a jail. Neither of these measures is appropriate
for measuring jail productivity.
Corrections—114
Program services. Program services such as academic and vocational training, recreation, counseling, religious services, and so forth are provided by some jails. While
they are not the only measure of jail output, they should be considered in preparing any
productivity index because they can consume scarce resources.
Jail surveys have found, however, that program services are not very important, at
least in terms of total resources expended. One survey found that less than 15 percent
of the Nation’s jails provided any type of adult education program. About half had
some type of work release program but only 6 percent of the inmates participated; and
recreation, while fairly common, usually took the form of unsupervised exercise. In
short, there are very few program services provided in the Nation’s jails, and there are
few employees to provide such services.55
Also, measurement of program service output raises numerous conceptual issues,
and data with which to make such calculations for the Nation’s jails are missing. Hence,
this measure cannot be used to develop a jail output measure.
Quality
Has the quality of local jails changed through time in such a way that it has affected
productivity? Unfortunately, there is little agreement on what constitutes jail quality
and how it should be measured. Generally, when discussing jail quality, the focus usually is on space per inmate, cleanliness of the facility, recreational opportunities, food
quality, and so forth.
Some might argue that jail quality has improved through time given court imposed
actions, court appointed monitors, and development of jail standards. Many States now
inspect local jails on a regular schedule and a few States have enforcement powers. On
the other hand, crowding is now an issue in many jails where it wasn’t a few years ago.
Even when there is agreement on what constitutes adequate quality, there is little data
on which to make an assessment.
Given the conceptual arguments, the absence of hard analytical evidence, and the
lack of data, the issue of quality will not be pursued further in this bulletin, except for
crowding.
Crowding—A special
concern
Crowding is a problem for many jails as it is for many prisons. The basic problem is
the same, too many inmates and not enough space, but the underlying conditions and
variables are quite different.
First, jail crowding is a recent phenomenon, at least for most institutions, while prison
crowding is a long-standing problem. Second, fluctuations in the number of prisoners
held in jails are much greater than for those held in prison. Jail prisoner fluctuations
require more reserve capacity to handle the peaks and valleys. Third, there is greater
variability in crowding from jail-to-jail than there is from prison-to-prison. This is the
result of the location, number, and function of jails as contrasted with State prisons.
State prisons are better able to balance their workload because of the long-term internment of prisoners and the greater number of facilities. In many local jurisdictions there
is only a single jail. Also, the large jails, which are located in the major metropolitan
areas, are generally filled, while those in rural areas often have excess capacity.
In 1983, jails in the United States operated at 85 percent capacity. Jails with a
capacity of 50 or fewer inmates had an occupancy rate of 52 percent; those with a
capacity of 50-249 had an occupancy rate of 86 percent; and those with a capacity of
1,000 or more had an occupancy rate of 121 percent.56
In 1988, jails in the U.S. operated at 101 percent capacity. However, jails with an
average daily population of 1,000 or more inmates operated at 126 percent of capacity.
A total of 404 jails were under court order in 1988 to limit the number of inmates they
held.57
In 1993, jail capacity was measured at 97 percent, a drop from previous years. Crowding at facilities with 1,000 or more inmates dropped to 111 percent of rated capacity.
But the statistics varied dramatically by State. Virginia’s jails were rated at 160 percent
of capacity while North Dakota’s held only 43 percent of their capacity.58
Corrections—115
Jail crowding generates the need for more staff, and exerts greater wear and tear on
facilities. It also fuels prisoner tensions. In short, it magnifies operating problems.59
Crowding also affects other parts of the criminal justice system and the community as a
whole. Early and emergency release of inmates is common. Nineteen States reported
early release of prisoners in 1985 because of overcrowding.
How does productivity change when more and more prisoners are crowded into a
jail? The primary evidence for crowding comes from studies of prisons. Available
research suggests that crowding of prisoners leads to more than a proportional increase
in the number of guards. Hence, the conclusion is that crowding probably reduces labor
productivity if it has any affect at all.60
In the case of jails, the available evidence is skimpy. One study found that as jail size
and density increased, the number of inmates per staff has also increased. The same
study also found an inverse relationship between inmate density and the number of
suicides.61
Since jail crowding is a recent phenomenon nationally, there is little information on
how it affects jail productivity. The available information does not indicate a deleterious affect on quality, however, and no adjustments will be made for jail crowding in the
jail productivity index.
Recidivism and other
outcome measures
The outcomes or results of jail operations, such as the recidivism rate, the number of
individuals appearing or failing to appear for trial, and community safety, fall outside
the scope of this study.
Inmates confined
This section presents an output index for local jails. The ideal measure would capture the work involved in admission and release processing, housekeeping and security
chores, and program services. But data limitations restrict the measure to a single,
proxy output—the number of inmates confined. Computation of even this simple measure is difficult, as several issues come into play.
First, there are two basic types of inmate measures, average daily population (ADP)
and one-time counts. Each has its strengths and weakness. The average daily population count is a better measure when examining annual resource expenditures because it
captures change through time in the number of inmates and it is not affected by extraordinary events that may affect a one-time or 1-day count. On the other hand, a one-time
count is helpful in capturing certain types of data, such as detention status, that are not
usually available for an ADP count. For long-term trend computations there should not
be a great difference between the two measures. The Bureau of Justice Statistics censuses of jail inmates in 1983 and 1993 show a 1-day count increase of 105.7 percent
while the ADP count registers a 104.9 percent increase.
The 1-day count is used here to measure jail output for two reasons. First, the ADP
is not available prior to 1983 and second, the employee counts used in the labor benchmark calculations are 1-day counts.
Regardless of the type of inmate count used, it is important to include all inmates
held, not just those under the jurisdiction of the government operating the jail. This is
because local jails often hold inmates for other governments, including Federal and
State governments. The National Sheriffs’ Association survey of 1981 showed half the
jails holding non-jurisdictional prisoners. The 1988 Census of Jails found that about 12
percent of all jail inmates were being held for other governments.62
The Federal Government has long used local government jails to house Federal inmates awaiting legal “disposition.” In 1987, 69 percent of the 85,348 individuals held
under the custody of the U.S. Marshals Service were housed in State and local facilities.63 State governments also rely on local governments to alleviate prison overcrowding. In 1988, 16 States reported housing 12,200 prisoners in local jails because of
crowding.64 For productivity measurement, it is important to focus on the total inmate
population when counting the number of inmates, not just those under the jurisdiction
of the holding government, because labor inputs support total jail operations.
Corrections—116
The output index calculated and presented here for local government jails is an
unweighted index of the number of inmates as of June 30 for the years for which reliable data are available: 1970, 1978, and 1982 through 1993.
The 1982-93 series reflects the annual estimates published by the Bureau of Justice
Statistics of the number of inmates as of June 30 each year. These data are based on a
sample of the Nation’s jails, except in 1983, 1988 and 1993 when they are based on a
complete census. The sample data reflect data on all jails with an inmate population of
100 or more and a stratified random sample of jurisdictions with an inmate population
of fewer than 100 (table 53).
Table 53. Local government jail inmate index,
1970, 1978, and 1982-93
(1970 = 100)
Year
Number of
inmates
Index
1970 ...................
160,812
100.0
1978 ...................
162,782
101.2
1982 ...................
1983 ...................
1984 ...................
1985 ...................
1986 ...................
1987 ...................
1988 ...................
1989 ...................
209,582
223,551
234,500
256,615
274,444
295,873
343,569
395,553
130.3
139.0
145.8
159.6
170.7
184.0
213.6
246.0
1990 ...................
1991 ...................
1992 ...................
1993 ...................
405,320
426,479
444,584
459,804
252.0
265.2
276.5
285.9
Average annual
percent change:
1970-93 ..............
1970-92 ..............
1970-78 ..............
1978-82 ..............
1982-92 ..............
1982-93 ..............
4.7
4.7
.2
6.5
7.8
7.4
The local jail output index shows a long-term average annual increase of 4.7 percent. Between 1970 and 1978 there was little change. But, beginning with the late
1970s and early 1980s there was rapid increase, and from 1982 through 1993 the average annual increase was 7.4 percent. By 1993, there were about 460,000 inmates in
3,300 local government operated facilities.
Labor inputs
There were almost 165,500 local government jail employees in 1993.65 Those States
with the greatest number of inmates—California, Texas, Florida, and New York—have
the greatest number of jail employees.
The number of employees is a relatively good indicator of the resources expended
on local jail operations because 70 to 80 percent of all jail operating expenditures go for
compensation as noted earlier. Furthermore, an index of jail employment seems to be a
reasonably good indicator of how jail resources change through time.
Custodial employees, or guards, comprise most jail employment, 71 percent in 1993.
The other major labor force occupational groups are clerical and maintenance (13 percent), professional and technical such as doctors, nurses, and psychologists (7 percent),
administrative (6 percent), and education and other (2 percent).66 This occupational
Corrections—117
breakdown has remained fairly constant over the 1970-93 period.
Most jail employees (94 percent in 1988) work full time. As with prisons, the criminal justice dimensions of the job, such as the requirement for staff security clearances,
for round-the-clock operations, and for specialized job training, dictate that full-time
employees be used in most instances. But the percent of employees who were full time
varies by occupation. For guards it was 96 percent and for clerical and maintenance
workers it was 92 percent. For professional and technical workers, however, it was 78
percent and for education it was 60 percent. The majority of jail employees (73 percent) are male.67
Two labor indexes, FTE employees and total employment, should be calculated for
government productivity calculations. Because of data limitations, only one, total employment, is calculated for jails. The two indexes usually move in concert, and this
should also be the case for jails because of the large number of full-time employees.
Calculating a jail labor index can be difficult. No national annual series of jail employment exists. Furthermore, there is a real question as to how many individuals work
in local jails. Jails, in contrast to prisons, have many employees who perform multiple
jobs, some outside the jail. In some of the smaller jurisdictions, for example, police
may perform jail duties. However, in the larger city jails, a separate department is
usually responsible for jail operations and the police are not directly involved.
Most inmates are held in county jails that are operated by an elected sheriff. In such
cases, deputy sheriffs or marshals work in the jails, but they may also be responsible for
court security, serving warrants, and even routine law enforcement activities like patrol
and answering calls for service. A 1990 survey of sheriff’s departments found that jail
responsibilities consumed at least a third of the employees’ work hours in 40 percent of
the departments. “About 87 percent of the departments performed some jail-related
work.”68
The local jail employment index presented here is derived from two basic sources.
The first is the local government correctional employment data series that is prepared
annually by the U.S. Bureau of Census. Jail employees comprise about two-thirds of
this series. The other basic source is the periodic survey sponsored by the Bureau of
Justice Statistics and its predecessor organizations. These surveys, which were conducted for 1970, 1978, 1983, 1988, and 1993, are used to benchmark the Bureau of
Census local government corrections employment series. The Bureau of Census and
Bureau of Justice Statistics data were used to calculate four separate jail employment
indexes: 1970-78, 1978-83, 1983-88, and 1988-93.
The 1970-78 index reflects the year-to-year change in local government correctional
employees. These data were benchmarked to the 1970 and 1978 Bureau of Justice
Statistics jail surveys with several minor adjustments. That is, adjustments were made
to the data so that the Census and Justice survey dates matched, and the jail employees
for Hawaii and Vermont were removed from the totals because inmates of these States
were not included in the outputs.
The 1978-83 index is not benchmarked. Rather, it reflects the year-to-year change
in the Census of Governments local government correctional employment series for
1978-83. This approach was taken because the Bureau of Justice Statistics jail employment data collection questions were modified in 1983, and there are questions as to how
these changes affected the respondents’ answers. This is the most questionable part of
the index.
The 1983-88 and 1988-93 local government correctional data were benchmarked to
Bureau of Justice Statistics local jail survey results. The benchmark data were taken
from Bureau of Justice Statistics sponsored censuses that were conducted as of June 30,
1983, 1988, and 1993. Each survey collected the number of full- and part-time personnel at work on the survey date. Payroll and non-payroll employees are included. Nonpayroll employees are persons who work in the jail but are paid by other organizations
such as departments of health and education.
Corrections—118
To calculate the 1970-93 jail employment index, the four employment indexes were
linked. The average annual increase in jail employment between 1970 and 1993 is 7.3
percent with increases in 22 of the 23 measured years (table 54).
Table 54. Local government jail labor index, 1970-93
Year
1970-78
index
1970 ..................
1971 ..................
1972 ..................
1973 ..................
1974 ..................
1975 ..................
1976 ..................
1977 ..................
1978 ..................
1979 ..................
100.0
104.9
115.0
126.7
135.2
146.8
151.2
157.0
156.8
1980 ..................
1981 ..................
1982 ..................
1983 ..................
1984 ..................
1985 ..................
1986 ..................
1987 ..................
1988 ..................
1989 ..................
1990 ..................
1991 ..................
1992 ..................
1993 ..................
1978-83
index
1983-88
index
1988-93
index
100.0
104.9
115.0
126.7
135.2
146.8
151.2
157.0
156.8
163.6
100.0
104.3
109.7
111.3
116.4
124.5
Linked
index
100.0
109.0
118.4
125.5
141.6
154.3
100.0
115.8
171.9
174.5
182.4
195.1
212.8
230.9
245.0
276.2
301.1
348.7
132.1
147.5
158.0
166.1
397.8
444.3
475.8
500.2
Average annual
percent change:
1970-92 .............
1970-93 .............
7.3
7.3
1970-78 .............
1978-83 .............
1983-88 .............
1988-93 .............
5.8
4.5
9.1
10.7
Productivity indexes
Local government jail productivity, as measured by output per employee, decreased
2.4 percent per year between 1970 and 1993. However, the period was marked by two
distinct trends. Between 1970 and 1978, the average annual decrease was 5.3 percent;
between 1978 and 1992, the decrease was 0.8 percent annually (table 55).
Corrections—119
Table 55. Local government jail labor productivity indexes, 1970-93
(1970 = 100)
Year
Output
Input
Productivity
100.0
100.0
130.3
139.0
145.8
159.6
170.7
184.0
213.6
246.0
100.0
104.9
115.0
126.7
135.2
146.8
151.2
157.0
156.8
163.6
171.9
174.5
182.4
195.1
212.8
230.9
245.0
276.2
301.1
348.7
1990 .................................
1991 .................................
1992 .................................
1993 .................................
252.0
265.2
276.5
285.9
397.8
444.3
475.8
500.2
63.4
59.7
58.1
57.2
Average annual
percent change:
1970-92 ............................
1970-93 ............................
4.7
4.7
7.3
7.3
-2.4
-2.4
1970-78 ............................
1978-82 ............................
1978-92 ............................
1982-87 ............................
1982-92 ............................
1987-93 ............................
.2
6.5
7.4
7.1
7.8
7.6
5.8
3.9
8.3
8.7
10.1
10.4
-5.3
2.6
-.8
-1.4
-2.0
-.7
1970 .................................
1971 .................................
1972 .................................
1973 .................................
1974 .................................
1975 .................................
1976 .................................
1977 .................................
1978 .................................
1979 .................................
1980 .................................
1981 .................................
1982 .................................
1983 .................................
1984 .................................
1985 .................................
1986 .................................
1987 .................................
1988 .................................
1989 .................................
Adult Probation and
Parole
101.2
64.6
71.5
71.2
68.5
69.1
69.7
66.6
70.9
70.5
Probation and parole is the third leg of the State and local government adult corrections stool, the other two legs being jails and prisons. Probation and parole is the
primary community-based function, and it is the largest of the three functions in terms
of the number of individuals handled.
Institutional setting
There were almost 3.4 million persons on probation or parole in 1992. Probation,
the most frequently used sentence in the criminal justice field, accounted for 2.8 million. The other 0.6 million individuals were on parole. Over the 1976-92 period, the
years for which comparable data are available, the number of persons on probation and
parole tripled. The average annual increase is 7.4 percent (table 56).
Corrections—120
Table 56. Number and index of probation and parole offenders under State and
local government supervision, 1976-92
(1976 = 100)
Year
Probation
Parole
Total
Index
1976 ...................
1977 ...................
1978 ...................
1979 ...................
923,064
967,567
1,020,770
1,037,944
156,194
159,447
160,556
191,767
1,079,258
1,127,014
1,181,326
1,229,711
100.0
104.4
109.5
113.9
1980 ...................
1981 ...................
1982 ...................
1983 ...................
1984 ...................
1985 ...................
1986 ...................
1987 ...................
1988 ...................
1989 ...................
1,072,728
1,179,223
1,308,130
1,532,721
1,688,597
1,913,334
2,057,484
2,186,776
2,325,398
2,463,019
196,786
203,418
203,331
230,115
250,138
283,139
308,763
343,902
387,145
435,381
1,269,514
1,382,641
1,511,461
1,762,836
1,938,735
2,196,473
2,366,247
2,530,678
2,712,543
2,898,400
117.6
128.1
140.0
163.3
179.6
203.5
219.2
234.5
251.3
268.6
1990 ...................
1991 ...................
1992 ...................
2,612,012
2,673,236
2,750,285
509,714
568,887
614,381
3,121,726
3,242,123
3,364,666
289.2
300.4
311.8
Average annual
percent change:
1976-92 ..............
7.1
8.9
7.4
7.4
SOURCE: Sourcebook of Criminal Justice Statistics, selected issues, and Probation and Parole, selected issues
Under both probation and parole the offender is released into, and supervised in, the
community. The principal difference between the two processes is how the procedure
is initiated. The court usually sets probation at the time of sentencing, although it may
follow a term of incarceration. Parole follows a prison or jail term and results from a
parole board decision or from mandatory conditional release by the courts. Supervision within the community can range from continuous monitoring through electronic
surveillance to little or no regular contact.69
Probation and parole law and practice varies by State. In some States, probation is
an executive branch function, in others it is a judiciary responsibility, and in still others
it is mixed. Juvenile probation and parole is usually a function of juvenile courts, an
activity that is often known as aftercare.
Probation can be a State or local government function. A 1976 study found that 32
States operated probation systems, 12 States assigned that responsibility to local government, and 6 had some type of combination.70 A 1988 study found a similar situation. The States assigning probation to local government often retained oversight responsibility for such functions as finance, setting of standards, and training. 71 The 1988
survey also found that 43 percent of the local offices provided probation services, 21
percent provided parole services, and 36 percent were responsible for both probation
and parole services.72
Most offenders released from prison are placed on parole.73 However, the percent
released into parole varies by State. In 1965, some States released all prisoners into
parole. But in recent years there has been a movement away from parole and toward
determinate sentencing. In 1977, 72 percent of the inmates released from prison were
released at the discretion of a parole board. In 1987 the figure was 41 percent. And at
least one State, Maine, eliminated all discretionary parole.74
Pardons of adult offenders are often included in discussions of probation and parole.
Data on this area are also commonly lumped together. But, in many cases, the data do
not indicate whether pardons are included or excluded. In any event, pardons are exCorrections—121
tremely limited, particularly when compared to probation and parole. Hence, it is an
issue that can safely be ignored in the calculations.
Definitions of each of these four functions—probation, parole, pardons, and aftercare—are presented in table 57.
Table 57. Definitions of probation, parole, pardon, and aftercare
Probation
is a sentence whereby the convicted person is released into the community rather than being incarcerated Subject to certain conditions
and restrictions imposed by the court, the person is subject to supervision by the probation authority.
Parole
is the release of a prisoner by the decision of a paroling authority
before the prisoner’s full sentence has been served. As with probation, the person released is subject to certain conditions and restrictions imposed by the paroling authority.
Pardon
is a popular term for clemency. Technically it is one type of clemency.
A pardon can be absolute or conditional.
Aftercare
is a commonly used term for juvenile parole. Juvenile parole and
aftercare are used interchangeably.
Probation and parole operations are specified in the Standard Industrial Classification Manual as Individual and Family Social Services, SIC Code 8322. The Manual
lists about 40 different types of operations, including parole and probation offices, under SIC 8322.75
Statistics are lacking as to what part of SIC 8322 resources go to support State and
local government parole and probation operations, and what part go to support other
activities covered by this SIC code.
The SIC 8322 code description does not specifically mention pardons, and so far as
is known, this function is not mentioned any place in the SIC Manual. The Bureau of
Census public employment definitions include pardons in the probation and parole category. The inclusion or exclusion of pardons should not have much of an affect on
resource, output, or productivity calculations as noted above. The function is just too
small.
Aftercare, or juvenile parole and/or probation, is sometimes included in general discussions of probation and parole. The SIC Manual does not explicitly mention juvenile
parole, probation, or aftercare. Nor does the Census of Governments documentation
assign it to a specific category. State treatment varies, and it is not obvious how specific
States handle their reporting.
Probation and parole work is commonly divided into two parts, investigation and
supervision. A 1967 study found that probation officers spent about 25 percent of their
time on investigation and 75 percent on supervision (i.e., case management).76 A Michigan report noted that 34 percent of the work units (time) were spent on investigations
while 66 percent were spent on supervision.77 However, a New York State study reported that the costs were roughly divided between investigation and supervision.78
Investigation involves examination of the background of the suspect, usually for a
judge but sometimes for a parole authority. The results of the investigation are used to
determine bail, set sentencing in case of conviction, set conditions for early release, set
prison classification, plan treatment, and plan parole.
A New York study reported four types of investigation: Prepleas, presentence, support, and juvenile delinquents. The time required per investigation ranged from 1 to 6
hours79 A Michigan report noted that of the investigative time, presentence investigation accounted for 64 percent, special 10 percent, preparole 3 percent, and other 23
percent.80 A California study found that presentence investigation (PSI’s) accounted
for half of probation officers’ time.81
For the country as a whole, the most common type of investigation is the presentence
investigation. In at least 22 States it is mandatory for all felon cases, and in 19 States it
Corrections—122
is required when probation is a potential sentence. In the remaining States it is discretionary. For misdemeanor offenders it is rarely completed. There has been a movement
in recent years to greater standardization of the PSI, and to cut back its scope and
length.82
Supervision oversees individuals on probation or parole to ensure that they adhere to
the terms of their court imposed sentence. The types and levels of supervision range
from constant electronic surveillance to intensive supervision where the probation officer sees the probationer or parolee several times each week to simple mail contact
once a month to no regular contact. Supervision usually means checking and verification but also can include referrals to education, job training, housing, and counseling.
“Only a small percentage of (the) probation population is being monitored, either by a
person or by an electronic device.”83
A New York State study presented statistics for three types of supervision of criminals: Intensive which required 3 to 4.5 hours per month of supervision; active which
required 1 to 1.5 hours each month; and special which required a half hour to threequarters of an hour of supervision each month. The study also found that the higher the
caseload the less time spent on the activity, and the less time spent with each client.84
The work of probation officers has changed over the past several decades. Today,
officers are less concerned with the provision of services such as counseling, and more
concerned with control, specifically, drug testing, curfew violations, employment verification, surveillance, and revocation procedures.85
Although, there has been some research and experimentation on the effectiveness of
parole and probation services over the past 30 years, there has been little discussion of
efficiency and productivity. The lack of interest carries over to data collection. Where
national statistics have been collected on the prison population for decades, it is only
recently that national parole and probation data have been collected and published.
Collection of statistics on the work of probation and parole units is ad hoc.86
Probation and parole is a disparate field with different institutional arrangements.
Furthermore, there has been great change in the work processes over the past several
decades, change that continues today. Because of this change it is difficult to collect
and array a consistent set of statistical data that can be used to measure labor productivity.
Outputs
This section considers the measurement of probation and parole outputs. As with
other government services, outputs are examined, not outcomes or impacts. The effectiveness of probation and parole operations, such as the recidivism rate, while important, falls outside the domain of productivity the way that the term is used here. Consequently, it is the work of State and local government probation and parole employees
that is of interest.
The following have been suggested as possible measures of adult probation and
parole output:
•
•
•
•
•
Number of parolees and probationers
Number of parolees and probationers separately counted and weighted
Number of parole and probation services differentiated by basic type
Number of parole and probation services differentiated by type and level of
supervision
Number of parole and probation work units
Each measure is briefly reviewed, the strengths and weaknesses discussed, and the
results summarized. The criteria used to evaluate the candidate output measures were
discussed in chapter 2.
Number of parolees and probationers. The first measure, and probably the most frequently cited measure of adult probation and parole output, is a count of the number of
offenders under supervision at a given point in time. National data are collected annually by the Bureau of Justice Statistics from which this index can be calculated.
Corrections—123
While this measure satisfies most of the output evaluation criteria presented in chapter 2, it has two serious defects. One, it does not measure the service provided; rather it
measures the number of individuals who might receive the service. Two, it does not
take into account the different labor units spent in producing the outputs. These two
deficiencies would not be a problem if each offender received the same type and level
of service, and this did not change through time. However, this is not the case. Although the statistics are skimpy, there are a number of individuals who are under supervision and counted, but who receive little or no service. And the type of service appears
to have changed through time. These are fatal flaws for this measure.
Number of parolees and probationers separately counted and weighted. A variation of
the first measure, this measure counts parole and probation separately, weights them,
and then combines them to create the output index. This measure assumes that the
service levels and unit labor requirements differ for probation and parole, a likely fact.
It also assumes that probation and parole services can be adequately measured by simply differentiating and weighting the two functions.
Limited evidence suggests that parole cases require higher supervision levels than
probation cases. One study found that 11 percent of the parole cases required intensive
supervision and 35 percent required maximum supervision. The study also noted that
parole officers had median caseloads of 65 cases each while probation officers carried
109 cases each. In the case of investigations, probation reported 129 studies per month
compared to 75 for parole. Whether the ratio between probation and parole changed
through time or if it changed at the same rate was not noted.87
This measure suffers from many of the same deficiencies of the preceding measure.
That is, within the individual functions, parole and probation, the types and levels of
service vary, and they change through time. Not everyone on probation and parole will
receive the same level of service. This candidate measure may be an improvement over
the first measure, but only marginally. It is unlikely to be satisfactory for measuring
probation and parole productivity output.
Number of parole and probation services differentiated by basic type of service. This
measure attempts to overcome the deficiencies noted in the first two measures by focusing on three basic types of service for parole and for probation. They are: Number of
intakes conducted; number of investigations made; and number of offenders supervised. Each should be weighted using the appropriate unit cost or labor weights and the
outputs combined to create the output index.
Conceptually, this measure is preferred to the first two measures that make no allowance for the different types of service provided. Whether it is sufficiently differentiated
to accurately track outputs is not known, and because data are lacking to compute this
measure, there is no way to evaluate this question.
Number of parole and probation services differentiated by type and level of supervision. This measure is the same as the previous one except that the level of supervision
is differentiated. The level of supervision can have a real impact on output per staff
member. One study noted that while there was no scientific basis for making a selection, three levels of supervision seem to be the norm in most departments of corrections.88
A study in Missouri in the early 1980s presented statistics on three types of supervision—intensive, regular and minimal—and four types of investigation—presentence
investigation (PSI), partial PSI, 30-Day, and other.89 The amount of work varied by
type of case and type of investigation. Also, both seemed to vary through time. The
amount of time devoted to each of these tasks varied greatly. Another article noted that
regular probation had as many as 300 cases per probation officer while intensive supervision had 33.90
Data collected by the Bureau of Justice Statistics show that between 1985 and 1992
the number of active probation cases increased by 42 percent, inactive 99 percent, those
Corrections—124
who absconded 78 percent, and those supervised out of State 39 percent. For parole,
the percentage increase was even larger: 119 percent for active, 217 percent for inactive, 179 percent for absconders, and 58 percent for out of State.91
Further differentiation of output, as recommended by this measure, would be helpful. The problem with calculating this measure, as with measures 3 and 5, is data availability.
Number of parole and probation work units. An even more carefully defined measure
of probation and parole outputs is the number of work units produced. This is essentially an activity-based measure. The study of New York probation work units noted
earlier identified multiple types of work units such as supervision of adult misdemeanants
and felons, children in need, juvenile delinquents, juvenile offenders, and persons who
do not support their families. But it did not distinguish between the different types of
supervision, nor did it address parole operations.92 To compute an index using work
units requires data for several dozen different outputs. While desirable analytically,
this approach is clearly infeasible for a national probation and parole output index.
Review. All of the measures discussed in this section are deficient in one way or the
other. For several, it would be impossible to collect some kinds of data; for others,
where local data are available, national data are not; and for others, where data exist the
measure is unsatisfactory conceptually because they do not capture the work of probation and parole organizations.
Labor inputs
Labor is a good measure of the resources used in probation and parole operations
because it accounts for such a large percent of the function’s operating expenditures.
While national statistics are lacking, labor apparently accounts for an even greater share
of the probation and parole operating costs than it does for prisons and jails. According
to a study of the New York City Division of Parole, personnel costs amounted to 84
percent of total parole costs in the mid-1970s.93 Another study, this one of two Arizona
counties, found that salary costs alone accounted for over 80 percent of total probation
operating costs.94 Given the labor-intensive nature of probation and parole work, these
statistics are quite understandable.
In 1990, there were 72,040 State and local government full-time equivalent employees in the probation, parole, and pardon field according to the Bureau of Justice Statistics. In 1971 there were 34,200. Thus, there has been a doubling of the number of FTE
employees over the 19 years for which data are available. But, as a percent of total
correctional employment, probation and parole has dropped from 23 percent in 1971 to
14 percent in 1990 (table 58).
From the limited data available, it appears that correctional employment and probation and parole employment increased at about the same rate during the early 1970s.
However, starting in the latter part of the 1970’s, probation and parole employment
lagged behind the rate of increase in overall correctional employment. This reflects the
dramatic increase in prison and jail employment.
As noted earlier, both State and local governments operate probation and parole
services. In 1990, there were about 15 percent more State employees than local employees (38,329 versus 33,365). Local government probation and parole employment
is predominantly county government employment; municipalities account for a small
part.95 In 1988, most, 96 percent, State and local government probation and parole
employees worked full time. 96
Corrections—125
Table 58. Comparison of probation and parole and total corrections full-time
equivalent employment, selected years
(1971=100)
Probation and parole
Total corrections
Year
1971 ...................
1972 ...................
1973 ...................
1974 ...................
1975 ...................
1976 ...................
1977 ...................
1978 ...................
1979 ...................
1980 ...................
1981 ...................
1982 ...................
1983 ...................
1984 ...................
1985 ...................
1986 ...................
1987 ...................
1988 ...................
1989 ...................
1990 ...................
1991 ...................
1992 ...................
Number
Index
34,200
37,200
39,500
46,000
100.0
108.8
115.5
134.5
51,301
150.0
57,158
167.1
63,893
186.8
72,040
210.6
Number
Index
151,000
160,000
172,000
185,000
193,000
206,000
218,000
222,000
236,000
100.0
106.0
113.9
122.5
127.8
136.4
144.4
147.0
156.3
249,000
259,000
277,000
297,000
325,000
350,0001
369,000
402,780
435,237
469,215
164.9
171.5
183.4
196.7
215.2
231.8
244.4
266.7
288.2
310.7
503,181
521,880
533,569
333.2
345.6
353.4
1
Estimated because FTEs were not calculated by the Bureau of the Census for 1985.
SOURCE: Probation and parole: Justice Expenditure and Employment in the U.S., selected issues, and Sourcebook of Criminal Justice Statistics, selected issues. Total Corrections: U.S. Bureau of Census, Public Employment, selected issues, Washington: Government Printing Office.
For government productivity measurement, the usual labor measures, as discussed
in chapter 2, are the total number of employees and the number of full-time-equivalent
employees. In the case of probation and parole it, both should be calculated, if possible, because the FTE numbers appear to be increasing at a more rapid rate.
There are many problems with the probation and parole employment data. First, the
information is spotty. For about half the years even the basic data are lacking. This
would not be a concern if good benchmark data were available, but they are not. For
most years data on the number of non-paid personnel are lacking (i.e., personnel who
work for other government agencies such as the courts). This apparently is a particular
problem for juvenile aftercare operations. For other years, data are available on State
employment but there are no data (or only partial data) on local employment. Another
problem is being able to relate employment with work. As noted in the discussion of
outputs, basic information about the work of employees is lacking. Do they work on
probation, parole, pardons, presentence investigations, or supervision of those released?
In short, the available employment data raise as many questions as they answer.
Conclusions
As noted earlier, it is not possible to compute a national probation and parole labor
productivity index because of the lack of data. There are problems with both inputs and
outputs, but the primary problem is the measurement of output. The only readily available national data are counts of the number of persons assigned to probation or parole,
data that say nothing about the work of probation and parole employees.
Corrections—126
Juvenile
Correctional
Institutions
Juvenile institutional operations are often set apart from other correctional operations, and sometimes are not even included in discussions of correctional institutions.
Their categorization varies by State. They are included in this bulletin because their
labor productivity measurement issues are similar to those of the adult correctional
institutions.
Institutional setting
There are over a thousand State and local government operated juvenile correctional
facilities, a number that has remained relatively steady since 1979. In 1991, the States
operated 47 percent of the 1,076 facilities and local authorities operated the rest. 97
Although State and local governments both operate juvenile facilities, the focus of
their operations is quite different. Local governments tend to focus on intake, screening, initial processing functions, and short-term detention. They handle about threequarters of all juvenile admissions to State and local government facilities. State governments, on the other hand, tend to focus on long-term care, and in 1991, they held 62
percent of all juveniles in State and local government facilities.98
Every State operated at least one juvenile facility in 1991. New York operated 66,
the most of any State. The greatest number of local government operations was in
California where there were 92. Thirteen States had no local government juvenile institutions.99
In 1991, State and local government institutions held almost 58,000 juveniles. In
1971, the first year for which comparable data are available, the figure was almost
55,000. Although the number of juveniles increased about 5 percent between 1971-91,
the number of facilities increased by almost 50 percent. As a result of the growth in the
number of facilities the number of residents per facility dropped. In 1971 there were 76
residents per facility; in 1991 there were 54.100
Cost data are lacking which would permit estimates of State and local government
juvenile correctional institutions as a percent of the total correctional population. State
data suggest that about 10 percent of State correctional operating budgets go to support
juvenile operations. In 1990 this was about $1.4 billion. A survey of county governments in 1988 found that about 15 percent of their corrections budget, or $315 million
was devoted to juveniles; this number cannot be extrapolated to all local governments
because of the differences in how juveniles are handled by local government.101 In
some States local governments have no responsibility but in others they have total responsibility.
In addition to government operated juvenile facilities, a number of private, nonprofit and for profit, juvenile operations exist. Indeed, there are almost twice as many
private facilities as public facilities (2,032 versus 1,076 in 1991). The number of private facilities has been increasing more rapidly than the number of public institutions.
Most States have more private facilities than public. In the early 1980s, Vermont used
only private facilities to house its juveniles. 102
The majority of juveniles, however, are still held in government operated facilities,
and the type of facilities operated by the two sectors is quite different. Governmentoperated facilities, which mostly operate in a closed environment, oversee the more
violent juveniles.103 Government held 80 percent of its juveniles in secure facilities in
1985 while private operators held only 16 percent. Private facilities, on average, held
fewer juveniles per facility (17 versus 47 for government) and kept juveniles longer
(126 days versus 41 days for government). The cost per juvenile per year was quite
similar in 1984 ($25,200 government versus $24,329 private).104
Juveniles are often classified and separated into delinquents, status offenders, and
non-offenders. Delinquents are juveniles who have been charged with or have been
convicted of a crime. This is the group that, for the most part, populates government
facilities. Status offenders are juveniles who are accused of committing or having committed an offense, which would not be an offense if committed by an adult. Examples
are school truancy, running away from home, ungovernability, and consumption of alcohol. Nonoffenders are juveniles who are not charged with any offense, but for lack of
Corrections—127
other alternatives end up in the juvenile correctional system. These include neglected
or abused children.105
There has been a concerted attempt through time, but especially over the past 20
years, to remove all non-delinquents from juvenile institutions. The 1974 Juvenile
Justice and Delinquency Prevention Act sought to remove those juveniles who were
neglected, dependent, abused, emotionally disturbed, mentally retarded, voluntarily admitted, and awaiting court disposition.106 Status offenders, those who were held for
committing non-criminal acts, such as truancy and running away from home were also
removed. From 1975 to 1985 the number of status offenders in public juvenile institutions was reduced by 49 percent. One result of the passage of the 1974 Act was to
increase the juvenile delinquent population, as a percent of the total, from 73 percent in
1975 to 93 percent in 1985. Another result was to reduce, at least temporarily, the
number of admissions to public juvenile institutions.107
Removal of juveniles who were confined for non-criminal acts has also shifted the
male-female ratio in public juvenile institutions. In 1975, females comprised 19 percent of the juvenile population, in 1985 they made up 14 percent, and in 1991 they were
11 percent.108
The age a person is classified as a juvenile varies by State. As of 1994, three States
set the upper age limit for juveniles at 15, eight States set it at 16, and the remaining set
it at 17. Except for Wyoming, the age was not changed between 1978 and 1994.109
Every State permitted juveniles to be charged as adults in certain cases.110 The average
age of juveniles held in State and local government facilities in 1983 was 15.4 years.
This is not an issue here because individual State definitions and statistics are used for
measurement purposes.
The productivity measurements discussed in this section relate to State and local
government juvenile institutions as specified by the Standard Industrial Classification
Manual, code 8361:
Establishments primarily engaged in the provision of residential social and
personal care for children, the aged, and special categories of persons with
some limits on ability for self-care, but where medical care is not a major
element. Included are establishments providing 24-hour year-round care
for children. Boarding schools providing elementary and secondary education are classified in Industry 8211. Establishments primarily engaged in
providing nursing and health-related personal care are classified in Industry Group 805.111
The Manual lists about 30 different types of facilities covered by this code ranging
from “Boy’s towns” to “Homes for destitute men and women.” The juvenile facilities
specifically mentioned are “Juvenile correctional homes, halfway homes for delinquents
and offenders, and training schools for delinquents.”
Statistics are lacking as to what part of SIC code 8361 resources go to support juvenile facilities, and what part go to support the other activities covered by this group.
However, it is assumed that juvenile operations are a small part of the operations covered by this SIC code. Private residential care is a large industry in the United States,
employing well over 500,000 individuals in 1991. Government juvenile facilities employed about 62,000 individuals in the same year.112
Employment in juvenile institutions is a small part of total employment in the overall
State and local government correctional field. In 1991, juvenile institutional employment accounted for about 10 percent of such employment. Although there has been an
increase in juvenile institutional employment over the past two decades, it has not kept
pace with the massive growth in employment in prisons and jails.113
Compared to many other State and local government functions, government employment and expenditures for juvenile operations is of minor importance. However,
there has been considerable interest in the field of juvenile corrections for many years.
Statistics were collected and published on reformatories in the United States in middle
Corrections—128
of the nineteenth century, and the 1870 report of the U.S. Commissioner of Education
included statistics on juvenile facilities. Data have been collected and published periodically by the Federal Government every since.114
Outputs
This section discusses the measures used to assess the output of juvenile institutions
in the United States, selects the measure to be used in the computations, and computes
and presents an index.
Output measures
There is considerable discussion of the juvenile correctional field, including proper
treatment modalities, recidivism rates, and outcomes. Nevertheless, there is little discussion of how to measure the output of juvenile institutions for productivity measurement purposes.
This section examines the work of State and local government juvenile detention
operations, such as the number of residents, number processed, the number released,
counseling provided, courses taught, and the like. The following measures have been
suggested in personal discussions as possible measures of juvenile institution output:
•
•
•
•
•
•
•
Juvenile residents
Juvenile residents by type of facility
Residents (juvenile and adult)
Average daily population (ADP)
Facilities
Admissions and discharges
Program services
Each suggested output is briefly reviewed, and its strengths and weaknesses discussed. The criteria used to evaluate the candidate output measures were listed in chapter 2.
Juvenile residents. Probably the most frequently cited and used measure of juvenile
institutional output is a count of the number of juveniles housed. The statistic is often
broken down by sex, age, and inmate status. This measure meets most of the output
evaluation criteria cited in chapter 2. It is simple, straightforward, and uses physical
counts; data are readily available to compute this output for a number of years. There is
no apparent research that discusses the relationship between this output and the resource units spent in its production.
Juvenile residents by type of facility. A variation of the count of juveniles is a count by
type of facility in which they are held. The six basic types—detention centers; shelters;
reception/diagnostic centers; training schools; ranches, forestry camps, or farms; and
half-way houses or group homes—are briefly described in table 59. The reason for
differentiating outputs by type of facility is that their unit resource requirements differ,
and there have been shifts through time in the growth of outputs by type of facility.
Corrections—129
Table 59. Types of basic juvenile facilities
Detention center
A short-term facility that provides custody in a physically
restricting environment pending or following adjudication,
pending disposition, placement, or transfer.
Shelter
A short-term facility that provides temporary care similar to
that of a detention center but in a physically unrestricted
environment.
Reception or diagnostic
center
A short-term facility that screens persons committed by
courts and assigns them to appropriate custodial facilities.
Training school
A long-term facility for adjudicated juvenile offenders typically under strict physical and staff control.
Ranch, forestry camp,
or farm
A long-term residential facility for persons whose behavior
does not require the strict confinement of a training school,
often allowing them greater contact with the community.
Halfway house or group
home
A long-term, nonconfining facility in which residents are allowed extensive access to community resources, such as
schooling, employment, health care, and cultural events.
SOURCE: Children in Custody, 1975-85, NCJ-114065, Washington, DC: U.S. Bureau of Justice Statistics, May 1989, p. 4
By differentiating the residents by type of facility some of the differences in program
services are captured. For example, facilities that care for juveniles on a long-term
basis, in a relatively open environment, have the lowest average unit labor requirements. There are two reasons for this: less need to process residents in and out of the
facilities and less need for security personnel to maintain discipline. Examples of such
facilities are ranches, camps, and farms and halfway houses and group homes. In the
latter case, direct costs are further reduced by the use of community schools, recreation
centers, and health clinics, that is, facilities that are operated by non-juvenile institutional staff.
This output measure should be superior to the first measure for productivity purposes because it differentiates and weights outputs by their unit labor requirements.
Like the first measure, it is simple, straightforward, and uses readily available data.
Residents (juvenile and adult). The first two measures reflect the number of juvenile
residents. A better measure of output for productivity measurement is a total resident
count, not just a juvenile count. Most residents of juvenile institutions are juveniles,
but a small percentage are adults. The percent has fluctuated from a high of 5.0 percent
in 1974 to a low of 1.7 percent in 1993.115
There are two reasons for housing adults in juvenile facilities. First, there are individuals who are confined as juveniles but age to majority while confined, and second,
there are juveniles who were tried and convicted as adults but are held in juvenile facilities for their own protection. For productivity measurement purposes all residents should
be counted since the resources support all residents, not just juveniles. So long as the
ratio between juveniles and all residents is reasonably constant through time it should
not matter which one is used for calculation of trends.
This measure, like the first two, is simple, straightforward, uses physical counts, and
summary data are readily available since 1971. What is lacking is information on the
type of resident by type of facility.
Corrections—130
Average Daily Population. Another commonly used measure of juvenile institutional
output is the average daily population (ADP) count. This is usually a better measure
than the 1-day count when assessing output over an extended period, such as a month or
year, because it captures the change through time in the number of residents. Also the
ADP is not affected by extraordinary events that may affect one-time counts. The day
or time of day that the count is taken may affect the one-time count. A Sunday count is
likely to be somewhat different from a Monday count.
A comparison of the ADP and 1-day count shows that in some years the ADP is
greater and in others the 1-day count is greater. Conceptually, the average daily count is
preferred, but because ADP is not readily available by type of facility, the 1-day census
count is often used. Like the preceding measures, the ADP is a physical count, often
used, and readily understood. There are questions concerning the calculation of the
average daily population measure, which were discussed in the State prison section.
Facilities. The number of facilities is sometimes mentioned as a measure of output.
While it is a readily available statistic, it is not a good measure of output for productivity measurement. Facility size varies and has changed through time, and productivity is
affected by many factors other than the number of facilities. While a facility count is
measurable, repetitive, accurate, readily collected, and readily understood it is not the
final service of the organization. It is not considered further as a measure of output.
Admissions and discharges. There is considerable turnover of the juvenile population.
The number of admissions has varied from 616,766 in 1971 to 527,759 in 1984 to
683,636 in 1990. The ratio between the number of admissions and the number of
residents has varied between 10 and 14 between 1971 and 1990. Admitting and discharging juveniles requires some government resources, but how much is not known.
The data needed to compute the unit resource weights for admissions and discharges
are lacking and, as a result, an output index cannot be calculated.
Program services. Program services such as academic education, vocational training,
recreation services, counseling, religious services, and the like are provided in most, if
not all, juvenile institutions. The type, number, and intensity of such services vary by
facility. Such activities should be considered in preparing any productivity index because they can consume considerable resources.
The problem in such a computation is how to measure these services and how to
include them. These are the same issues discussed in the section on prisons and jails.
It is difficult to measure government program services in the best of circumstances. In
the case of juvenile institutional education, one might measure the number of classes,
the number of participants, and the number of graduates. But as with most educational
services, the question is what is the correct measure?
In the case of juvenile institutions the conceptual questions are moot because there
are no national output data on any of the program services. The only readily available
data are staff counts, which are inputs, not outputs.
Program service resources are partially captured when output is differentiated by
type of facility and weighted. For example, halfway houses or group homes use fewer
resources than training schools, and part of the difference is due to the program services
provided by the training schools.
Quality and crowding
This section briefly considers a number of other factors, such as quality and crowding, that might affect productivity output calculations. Although it is easy to discuss
such considerations and their potential effect on productivity in the abstract, there is
little agreement on what constitutes quality and how it should be measured. Generally,
when individuals discuss juvenile detention quality they focus on space per resident,
cleanliness, program availability, food quality, and so forth. No research was found on
how these factors affect productivity.
Institutional crowding is a special concern for juvenile institutions as it is for the
Corrections—131
Nation’s jails and prisons. However, the problem appears to be less acute for juvenile
facilities. The design capacity for juvenile institutions in 1985 was 56,895 while the
average daily population was 47,496 and the Census-day count was 51,402.116 That is,
there was almost 20 percent excess capacity for the ADP and 11 percent for the Censusday count. However, these statistics are overall averages and do not take into account
peak periods and local area distribution. In 1985, about 17 percent of juvenile facilities
were operating in excess of their design capacity; in 1971 the figure was 16 percent. In
1985, 36 percent of the juveniles were held in crowded facilities. In other words, a few
large facilities were responsible for most of the crowding.
What effect does juvenile facility crowding have on juvenile institutional productivity? No information was found that addressed this subject. The only information on
correctional institutional crowding comes from studies of prisons and jails. Some individuals have argued that crowding leads to increases in labor productivity.117 However,
most of the evidence suggests the opposite, that crowding has a negative affect on productivity.118 As with other correctional operations, the courts have intervened to ensure
minimal standards of juvenile operation. The data in this bulletin are not calculated to
adjust for juvenile crowding
Recidivism and other
outcome measures
The outcomes or results of juvenile institutional operations, such as the recidivism
rate, juveniles returned to their families, the number of residents moving on to a life of
crime, and the like, fall outside the domain of productivity measurement the way the
term is used here. This section deals with the work of State and local government
juvenile detention operations, such as the number of residents, number processed, counseling provided, courses taught, and the like. Hence, as with the other services discussed here, no attempt will be made to measure productivity outcomes.
The output series
This section presents an output index for juvenile detention institutions. Based on a
review of the available research and data, the following measure is proposed:
The number of residents differentiated by facility type: Detention centers; shelters; reception/diagnostic centers; training schools; ranches, forestry camps
and farms; and half-way houses and group homes.
Several other factors should also be included, such as the number of program services and the number of admissions and discharges, but data are lacking with which to
make such calculations.
National data to construct the index are taken from the U.S. Department of Justice’s
Office of Juvenile Justice and Delinquency Prevention (OJJDP) statistical collection
efforts. Data for 1971, 1973, 1974, 1975, 1977, 1979, 1983, 1985, 1987, 1989, 1991,
and 1993 are taken directly from OJJDP’s Children in Custody series. Data for the
missing years are estimated.
Outputs have been calculated for six different facility types (table 60). Each reflects
the number of juveniles from 1971-87 and the number of residents from 1987-93. The
preferred measure is the number of residents. But these data are not readily available
prior to 1987, thus, the two different counts are used and linked in 1987.
Table 60 shows rapid increases for halfway houses and group homes and moderate
increases for detention centers and shelters. The average annual increase in growth of
halfway houses and group homes was 5.1 percent with increases in about half the years.
The long-term growth in detention centers was 2.6 percent, but there are two distinct
periods of change. From 1971-77, the number of juveniles declined but from 1977-93
the number increased. The average annual increase from 1977-93 was 4.6 percent. While
the number of shelter residents increased substantially between 1971 and 1993 the growth
has been erratic. The reasons behind the abrupt changes between 1975-79 and 1984-87
are not known. Output declined for reception centers, training schools, and ranches,
camps, and farms over the 1971-93 period. Training schools, the largest facility type in
terms of the number of residents, registered a slight decline. The index declined from
Corrections—132
1971 through 1979. Since then, the index has increased at a 1.3-percent average annual
rate. Reception centers also showed a steady decrease through 1979, and have increased
since then. This is not surprising because they are the first step for most juveniles
moving on to a training center. Ranches, camps, and farms have registered a fairly
steady average annual decrease, 0.7 percent, throughout the period.
Table 60. Juvenile institutional output indexes by type of institution, 1971-93
(1971 = 100)
Reception Training
school
center
Ranch
farm
Halfway
house
100.0
88.9
77.7
74.7
78.7
74.0
69.4
68.8
68.2
100.0
95.3
90.6
95.6
98.4
98.5
98.6
93.7
88.8
100.0
132.3
164.6
177.5
218.1
256.9
295.8
295.6
295.4
55.0
59.5
64.1
68.6
67.4
66.2
66.5
66.7
64.6
62.4
69.6
71.0
72.4
73.7
73.7
73.7
76.6
79.5
80.3
81.1
90.7
92.6
94.5
96.4
95.1
93.7
88.1
82.5
82.6
82.7
306.4
317.4
328.4
339.5
329.9
320.3
346.2
372.0
370.6
369.2
185.3
191.1
182.6
174.2
70.1
77.9
80.4
82.9
82.0
83.0
82.5
82.0
78.3
73.9
80.1
86.3
364.7
360.3
329.0
297.7
2.6
-0.8
-0.9
-0.7
5.1
Year
Detention
center
Shelter
1971 .................
1972 .................
1973 .................
1974 .................
1975 .................
1976 .................
1977 .................
1978 .................
1979 .................
100.0
95.8
91.6
93.6
94.2
89.5
84.8
87.8
90.8
100.0
76.4
52.8
50.0
55.6
135.3
215.0
181.7
148.3
100.0
90.3
80.5
63.9
66.7
67.7
68.7
59.6
50.5
1980 .................
1981 .................
1982 .................
1983 .................
1984 .................
1985 .................
1986 .................
1987 .................
1988 .................
1989 .................
95.8
100.8
105.9
110.9
114.0
117.0
127.1
137.2
145.2
153.1
148.1
147.8
147.5
147.2
185.4
223.6
216.0
208.3
193.9
179.4
1990 .................
1991 .................
1992 .................
1993 .................
157.2
161.3
168.3
175.3
Average annual
rate of change:
1971-93 ............
2.6
A weighted output index was also calculated for all juvenile facilities. That is, the
six different types of facility outputs were weighted with their unit labor weights in
1971, 1977, 1983, and 1987, and combined to produce a single index. The base years
reflect those years closest to the base years that BLS used in its government productivity estimates (i.e., 1972, 1977, 1982, and 1987). Unit labor requirements for 1971,
1977, and 1987 were computed directly from the data collected and published by the
Department of Justice. Unit labor weights for 1983 were estimated using 1979 and
1987 data.
The results of the weighted (differentiated) output calculation shows that output was
relatively flat over the entire 1971-93 period, dropping during the early years and rising
thereafter. From 1971 through 1979, the average annual decrease was 3.1 percent. From
1979 through 1993, the average annual increase was 2.3 percent; there has been an
increase every year since 1979. The overall average annual increase between 1971-93
is 0.3 percent (table 61).
An unweighted index was also calculated. That is, the number of residents was combined and the yearly totals indexed. The results of this calculation show an average
annual increase between 1971-93 of 0.3 percent, the same as the weighted index. The
similarity in the indexes reflects the dominant role of the training schools and detention
centers, which account for over 80 percent of the residents.
Corrections—133
Table 61. Comparison of undifferentiated and differentiated
juvenile institutional output indexes, 1971-93
(1971=100)
Year
Undifferentiated
index
Differentiated
index
1971 .........................
1972 .........................
1973 .........................
1974 .........................
1975 .........................
1976 .........................
1977 .........................
1978 .........................
1979 .........................
100.0
91.7
83.5
82.1
85.8
83.2
80.6
79.8
79.0
100.0
91.5
83.0
81.2
84.6
81.8
79.0
78.5
78.0
1980 .........................
1981 .........................
1982 .........................
1983 .........................
1984 .........................
1985 .........................
1986 .........................
1987 .........................
1988 .........................
1989 .........................
81.5
84.0
86.5
89.0
89.6
90.1
93.9
97.8
99.7
101.6
80.6
83.3
85.9
88.6
89.4
90.3
94.4
98.6
100.6
102.6
1990 .........................
1991 .........................
1992 .........................
1993 .........................
102.9
104.2
105.5
106.8
104.2
105.7
106.8
107.9
Average annual
rate of change:
1971-92 ....................
1971-93 ....................
0.2
.3
0.3
.3
1971-79 ....................
1979-93 ....................
-2.9
2.2
-3.1
2.3
Labor inputs
There were about 62,000 State and local government juvenile facility employees in
1991; 58 percent were State employees and 42 percent were local. Of the total, roughly
53,000 were full time and 9,000 were part time. In addition, there were about 3,600
volunteers.119
The division between full-time and part-time employees varies by type of facility.
Employees in State run training facilities, for example, are mostly full time (94 percent). On the other hand, less than 67 percent of the local government half-way house
employees were full time.120
As with the other correctional fields, labor is the primary resource input into juvenile
institutional operations. In 1975, wages and salaries accounted for about 75 percent of
all operating expenditures.121 Adding employee benefits to these statistics would raise
the figure to well above 80 percent.
The total number of employees and the number of full-time equivalent employees
are the statistics normally used to calculate government productivity employment indexes (chapter 2). Because of data concerns, neither index is computed for juvenile
institutions. Instead, the total number of full-time employees is used. Data since 1987
suggest a fairly constant ratio between total and full-time employment. Prior to 1987,
this ratio is unstable and there are serious concerns about some of the part-time employment data.
The juvenile institution labor index is constructed in two parts and linked. For 1971Corrections—134
87, the index reflects the number of full-time payroll employees, full-time non-payroll,
and full-time volunteers.122 For 1987-93, the statistics reflect full-time payroll and nonpayroll staff only; volunteers are excluded. The two data series are linked in 1987 to
create the 1971-93 index.
Juvenile institutional employment data were collected for 1971, 1973, 1974, 1975,
1977, 1979, 1983, 1987, 1989, 1991, and 1993. Data for the intervening years were
estimated. The survey dates are the same as for the outputs.
The results of these calculations show that the average annual increase in juvenile
institutional labor was 1.3 percent between 1971-93 with an increase every year between 1973 and 1991. The increases have been steady and modest throughout the
period (table 62).
Table 62. Index of full-time employees in State and local government
juvenile institutions, 1971-93
(1971 and 1987 = 100)
Year
1971-87
index
1987-93
index
Labor
index
1971 .........................
1972 .........................
1973 .........................
1974 .........................
1975 .........................
1976 .........................
1977 .........................
1978 .........................
1979 .........................
100.0
99.6
99.2
99.7
104.1
106.9
109.6
110.9
112.2
100.0
99.6
99.2
99.7
104.1
106.9
109.6
110.9
112.2
1980 .........................
1981 .........................
1982 .........................
1983 .........................
1984 .........................
1985 .........................
1986 .........................
1987 .........................
1988 .........................
1989 .........................
112.9
113.5
114.2
114.8
117.1
119.3
121.6
123.8
100.0
101.4
102.8
112.9
113.5
114.2
114.8
117.1
119.3
121.6
123.8
125.6
127.3
105.7
108.6
108.3
108.0
130.9
134.5
134.1
133.7
1990 .........................
1991 .........................
1992 .........................
1993 .........................
Average annual
percent change:
1971-92 ....................
1971-93 ....................
1.4
1.3
Productivity indexes
Although labor inputs increased at a fairly steady pace for most of the measured
period, long-term outputs were fairly stable. The result was a drop in labor productivity
of 1.0 percent per year between 1971 and 1993. However, there were two very distinct
segments during this period. During the initial years (1971-79) labor productivity
dropped 4.5 percent annually. Indeed, every year from 1971-79 shows a decline in
productivity. But from 1979-93, productivity increased at an average annual rate of 1.1
percent (table 63).
In terms of absolute values, there were about 1.2 residents for each full-time (not
FTE) staff member in 1993. But this ratio varies by type of facility, ranging from 1.5
residents to each full-time staff for ranches, camps, and farms to .8 for shelters. The
ratio for training schools, which account for about 50 percent of those held in
juvenile facilities, is the same as the overall average.123
Corrections—135
Table 63. State and local government juvenile institution productivity
indexes, 1971-93
(1971 = 100)
Output
index
Labor
index
Productivity
index
1971 ...........................
1972 ...........................
1973 ...........................
1974 ...........................
1975 ...........................
1976 ...........................
1977 ...........................
1978 ...........................
1979 ...........................
100.0
91.5
83.0
81.2
84.6
81.8
79.0
78.5
78.0
100.0
99.6
99.2
99.7
104.1
106.9
109.6
110.9
112.2
100.0
91.8
83.6
81.4
81.3
76.6
72.1
70.8
69.5
1980 ...........................
1981 ...........................
1982 ...........................
1983 ...........................
1984 ...........................
1985 ...........................
1986 ...........................
1987 ...........................
1988 ...........................
1989 ...........................
80.6
83.3
85.9
88.6
89.4
90.3
94.4
98.6
100.6
102.6
112.9
113.5
114.2
114.8
117.1
119.3
121.6
123.8
125.6
127.3
71.4
73.3
75.2
77.1
76.4
75.7
77.7
79.6
80.1
80.6
1990 ...........................
1991 ...........................
1992 ...........................
1993 ...........................
104.2
105.7
106.8
107.9
130.9
134.5
134.1
133.7
79.6
78.6
79.6
80.7
Average annual
percent change:
1971-92 ......................
1971-93 ......................
1971-79 ......................
1979-92 ......................
1979-93 ......................
0.3
.3
-3.1
2.5
2.3
1.4
1.3
1.5
1.4
1.3
-1.1
-1.0
-4.5
1.1
1.1
1971-77 ......................
1977-83 ......................
1983-87 ......................
1987-92 ......................
1987-93 ......................
-3.8
1.9
2.7
1.6
1.5
1.5
.8
1.9
1.6
1.3
-5.3
1.1
.8
0
.2
Year
SOURCE: Tables 61 and 62
Summary and
Conclusions
Corrections—136
This chapter discussed the measurement of productivity indexes of State and local
government correctional services. Four were examined and average labor productivity
indexes were calculated for three—prisons, jails, and juvenile correctional facilities.
Probation and parole, the fourth area, is a very different operation, and could not be
measured because of lack of adequate output data.
Corrections is a general government service. All States, and many local governments, are involved in the incarceration of men, women, and young people. Although a
few governments contract with private firms to operate parts of their correction services—so called privatization—the function remains primarily a government one. An
exception is juvenile institutions where there is a large, active non-profit industry. In
this chapter, only government-operated facilities were measured.
State and local government spent almost $29 billion in fiscal 1992 on correctional
services (table 44). Capital expenditures accounted for about 15 percent, operations
about 85 percent. Because more than three-quarters of all operational expenditures are
devoted to labor compensation, its cost is a good indicator of the amount of resources
devoted to correctional operations.
There were over 540,000 State and local government correctional employees in
1992.124 The average annual increase in employment between 1967 and 1992 is over 6
percent.
Correctional employment is full-time employment. In 1992, 96 percent of all correctional employees worked full time.125 The nature of the work, the specialized training required to carry out the work, and the security background investigations required
of all employees, dictate the use of full-time employees.
The States employ 64 percent of all correctional employees. Local government,
primarily counties, employs the rest. This division reflects State operation of prisons,
and local government operation of most jails. State and local governments divide probation and parole and juvenile delinquency operations with the assignment depending
on State law.
Prisons employ the largest number of correctional workers, about half of the total,
jails follow with one quarter, and probation and parole and juvenile facilities employee
the rest.
State and local correctional operations are one of the fastest growing government
services. In response to community concerns, new laws have been passed, determinant
sentencing has been implemented, and, in some cases, parole has been abolished. One
result of these actions has been an increase in output for each of the three services and
large increases for prisons and jails.
The increase in output has generated additional demand for labor. But there have
been other pressures, too, probably the most important being court rulings. The courts
have taken an active role in the operation of many State and local government correctional institutions, including specification of staffing requirements. This has contributed to the larger proportional increase in labor compared to output. This, in turn, has
led to decreases in labor productivity as discussed below.
State adult institutions
Prison output and employment both increased rapidly during the measured period.
Output grew 7.8 percent per year between 1973 and 1992, and the number of inmates
increased every year (table 52). By 1992, there were over 750,000 inmates in State
operated facilities. Every State shows large increases in the number of inmates since
1973, and nine States show average annual increases of 10 percent or more.
State adult institutional productivity showed little change between 1973 and 1992,
but there were two slightly different trends. From 1973 through 1982, productivity
increased 1.3 percent annually but between 1982-92 it dropped 1.0 percent per year
(chart 7).
Chart 7. State government adult correctional institutions
output, labor, and productivity indexes, 1973-92
Index, 1973=100
450
450
Output
Productivity
Labor
400
350
400
350
300
300
250
250
200
200
150
150
100
100
50
50
0
0
1973
1977
1981
1985
1989
Corrections—137
Local jails
Local jails also registered increasing output and labor and decreasing labor productivity over the measured period, 1970-92. The average annual increase in output was
4.7 percent from 1970 through 1992, but there were two very distinct trends. From
1970 through 1978, output was fairly flat increasing 0.2 percent annually. From 1978
through 1992, the average annual increase was 7.4 percent, and every year from 1982
registered an increase (table 55).
Comparison of jail and prison statistics show that during the 1980s, jail and prison
output increased very rapidly. This is in marked contrast to the 1970s when jail output
was practically flat and prisons increased rapidly. One reason for the increase in jail
output in the 1980s is the increase in the number of State inmates being held in local
jails because of overcrowding in State facilities. In 1988, 8 percent of the local jail
population was being held for State authorities, mostly because of overcrowding in
State prisons.126
Labor input grew at about the same rate as output. In response to the increase in the
number of prisoners, prison crowding, more militant prisoners and court actions that
dictated additional staffing, employment increased every year between 1973 and 1992.
There were over 275,000 employees in 1992, an average annual increase of 7.8 percent
between 1973 and 1992.
Like prisons, jail labor increased rapidly and many of the same events that drove the
increase in prisons propelled the increase in jails: Crowding, more militant prisoners,
and court decrees. There were an estimated 157,500 local government jail employees
in 1992, an average annual increase of 7.3 percent between 1970-92.
One result of this increase was a drop in jail labor productivity. For 1970-92 productivity declined 2.4 percent annually; the average annual decline between 1982 and 1992
was 2.0 percent (chart 8).
Chart 8. Local government jail output, labor, and productivity
indexes, 1982-92
Index, 1982=100
300
250
300
Labor
Productivity
250
Output
200
200
150
150
100
100
50
50
0
1982
0
1984
1986
1988
1990
1992
Juvenile institutions
Over the long term, juvenile institutional labor productivity declined as it did for
prisons and jails. But, there were two very different trends in the case of juvenile
institutions. Between 1971 and 1979, productivity dropped at an average annual rate of
4.5 percent, but between 1979 and 1992, it turned around and increased 1.1 percent per
year (table 63). These trends are driven by the change in output. Output decreased
from 1971 through 1979, but increased from 1979 onward. The average annual change
for the two periods was -3.1 and 2.5 percent, respectively. Labor input increased at a
fairly constant rate after 1974. By 1992 there were about 62,000 employees employed
by juvenile institutions (chart 9).
Corrections—138
Chart 9. State and local government juvenile institution
output, labor, and productivity indexes, 1971-92
Index, 1971=100
140
120
100
80
60
40
20
0
1971
Comparison of the
indexes
Output
1976
Productivity
1981
Labor
1986
140
120
100
80
60
40
20
0
1991
Comparison of average labor productivity for prisons, jails, and juvenile facilities
for the years for which comparable data are available (1982-92), show decreases for
prisons (1.0 percent) and jails (2.0 percent), and a small increase for juvenile facilities
(0.6 percent). The long-term change varies depending on the period examined, as noted
in chart 10.
Chart 10. Labor productivity indexes for State and local
government prisons, jails and juvenile institutions,
1971-92
Index, 1992=100
140
120
100
80
60
40
20
0
1971
Prison
1976
Jail
1981
Juvenile
1986
140
120
100
80
60
40
20
0
1991
Overall averages
Three functions—prisons, jails, and juvenile institutions—have been combined to
produce a summary correctional institutional productivity index for 1973-92. The three
functions, while different, share many of the same problems and operational concerns.
Each provides food and shelter for offenders, each processes offenders in and out of the
system, each provides security, and each provides some treatment and program services.
New output and labor indexes were generated to compute the correctional institution
productivity index. Data for these indexes cover prison and juvenile institutions for the
entire 1973-92 period, and jails for 1982-92. The covered periods are dictated by data
availability. The output index was calculated by combining the outputs of the three
services using labor weights for the base years of 1973, 1977, 1982, and 1987. The
labor index uses the same benchmark years and simply reflects the employment levels
of the three services. The productivity index was calculated by dividing the output
index by the input index.
There were rapid increases in output and labor input between 1973 and 1992. The
average annual increase in output is 5.0 percent with increases every year. Labor also
increased every year. The long-term average annual increase in labor was 6.6 percent.
Corrections—139
The summary correctional labor productivity index decreased at an average annual
rate of 1.5 percent between 1973-92. However, the rate varies by the period examined.
From 1973-82 productivity increased at an average annual rate of 0.5 percent. But,
beginning in 1982, there was a decided decline with decreases every year from 1982.
The average annual decline over the 1982-92 period was 3.3 percent. (See chart 11 and
table 64.)
Chart 11. State and local government combined correctional
institutions output, labor, and productivity indexes, 1973-92
Index, 1973=100
400
400
Output
Productivity
Labor
300
300
200
200
100
100
0
0
1973
1977
1981
1985
1989
Table 64. Indexes of State and local government correctional
institution output, input, and productivity, 1973-92
(1973 = 100)
Output
Labor
Productivity
1973 ...................
1974 ...................
1975 ...................
1976 ...................
1977 ...................
1978 ...................
1979 ...................
100.0
104.4
113.2
118.8
121.8
125.0
128.4
100.0
104.5
109.5
115.4
119.9
123.2
131.5
100.0
99.9
103.4
103.0
101.5
101.5
97.7
1980 ...................
1981 ...................
1982 ...................
1983 ...................
1984 ...................
1985 ...................
1986 ...................
1987 ...................
1988 ...................
1989 ...................
134.7
148.1
163.3
168.9
173.3
182.0
192.0
199.7
213.2
228.3
134.2
145.4
155.8
167.2
182.8
195.5
207.1
225.7
250.3
276.6
100.3
101.9
104.8
101.1
94.8
93.1
92.7
88.4
85.2
82.5
1990 ...................
1991 ...................
1992 ...................
234.8
242.0
252.3
303.8
322.5
337.4
77.3
75.0
74.8
Average annual
rate of change:
1973-92 ..............
5.0
6.6
-1.5
1973-77 ..............
1977-82 ..............
1982-87 ..............
1987-92 ..............
5.0
6.0
4.1
4.8
4.6
5.4
7.7
8.4
.4
.6
-3.3
-3.3
Year
Corrections—140
Endnotes
1
U.S. Bureau of the Census, Census of Governments (1987), Vol. 4, No. 5, Compendium of
Government Finances, Washington: U.S. Government Printing Office, 1990, p. A-2.
2
U.S. Bureau of the Census, Census of Governments, 1967, Vol. 3, No. 2, Compendium of
Public Employment, Washington: U.S. Government Printing Office, 1969, p. 21, and U.S. Bureau of the Census, Public Employment: 1992, Series GE/92-1, Washington: U.S. Government
Printing Office, 1994, p. 5.
3
There were another 156,000 offenders under Federal control or supervision. Tracy L. Snell,
Correctional Populations in the United States, 1992, NCJ-146413, Washington: U.S. Bureau of
Justice Statistics, January 1995, p. 5.
4
U.S. Bureau of Justice Statistics, Census of State and Federal Correctional Facilities, 1990,
NCJ-137003, Washington: U.S. Government Printing Office, May, 1992, pp. 17 and 18.
5
U.S. Bureau of the Census, Public Employment: 1990, Series GE-90-1, Washington: U.S.
Government Printing Office, 1991, p. 5; U.S. Bureau of the Census, Government Finances:
1989-90, Series GF/90-5, Washington: U.S. Government Printing Office, 1991, p. 11; and U.S.
Bureau of Justice Statistics, Census of State and Federal Correctional Facilities, 1990, NCJ137003, Washington: U.S. Government Printing Office, May, 1992, pp. 14 and 17.
6
U.S. Bureau of Justice Statistics, Census of State and Federal Correctional Facilities, 1990,
NCJ-137003, Washington: U.S. Government Printing Office, May, 1992, pp. 2, 6, and 17.
7
Douglas McDonald, The Price of Punishment: Public Spending for Corrections in New York,
Boulder, Colorado: Westview Press, 1980, p. 17.
8
“Security Staffing in Adult Facilities of the Virginia Department of Corrections,” Richmond:
Virginia Department of Planning and Budget, December 1985, p. I-4.
9
“State and Federal Prisons: Factors That Affect Construction and Operations Costs,” Washington: U.S. General Accounting Office, May 1992, p. 10.
10
U.S. Bureau of Justice Statistics, Census of State and Federal Correctional Facilities, 1990,
NCJ-137003, Washington: U.S. Government Printing Office, May 1992, p. 10.
11
1984 Census of State Adult Correctional Facilities, NCJ-105585, Washington: U.S. Bureau of
Justice Statistics, August, 1987, pp. 7-10.
12
U.S. Bureau of Justice Statistics, Census of State and Federal Correctional Facilities, 1990,
NCJ-137003, Washington: U.S. Government Printing Office, May, 1992, pp. 1 and 20.
13
Robert M. Carter, et. al., Correctional Institutions, New York: Harper and Row Publishers,
1985, p. 102.
14
U.S. Bureau of Justice Statistics, Census of State and Federal Correctional Facilities, 1990,
NCJ-137003, Washington: U.S. Government Printing Office, May 1992, p. 7.
15
U.S. Office of Management and Budget, Standard Industrial Classification Manual. Washington: U.S. Government Printing Office, 1987, pp. 409-10.
16
U.S. Office of Management and Budget, Standard Industrial Classification Manual. Washington: U.S. Government Printing Office, 1987, p. 395.
17
The District of Columbia is a Federal city, but for statistical purposes it is treated as if it were
a State.
18
Marc Holzer, Constance Zalk and Jerome Pasichow, “Corrections,” in George J. Washnis,
Productivity Improvement Handbook for State and Local Government, New York: John Wiley,
1980, pp. 1003-31; and Louis H. Blair, et. al., Monitoring the Impacts of Prison and Parole
Services: An Initial Examination, Washington: Urban Institute, July, 1977.
19
Michael Block and Thomas Ulen, “Cost Functions for Correctional Institutions,” Charles M.
Gray (ed.), The Costs of Crime, Beverly Hills, CA.: SAGE Publications, 1979 and Paul Schmidt
and Ann D. Witte, An Economic Analysis of Crime and Justice, New York: Academic Press,
1984.
Corrections—141
20
Judith C. Hackett and others, “Contracting for the Operation of Prisons and Jails,” Washington: U.S. National Institute of Justice, June, 1987, p. 5.
21
. National Planning Association, The National Manpower Survey of the Criminal Justice System, Washington: U.S. National Institute of Law Enforcement and Criminal Justice, September,
1978, p. 16.
22
. Douglas McDonald, The Price of Punishment: Public Spending for Corrections in New York,
Boulder, Colorado: Westview Press, 1980, p. 26.
23
U.S. Bureau of Justice Statistics, Census of State and Federal Correctional Facilities, 1990,
NCJ-137003, Washington: U.S. Government Printing Office, May 1992, p. 15.
24
The 1979 survey found that 62 percent were custodial. The reason for the difference between
1979 and the other years is not known. It could be data error or a change in data classification;
it is unlikely that it is a real change.
25
U.S. Bureau of Justice Statistics, Report to the Nation on Crime and Justice, NCJ-105506,
Washington: U.S. Government Printing Office, March 1988, p. 107.
26
1984 Census of State Adult Correctional Facilities, NCJ-105585, Washington: U.S. Bureau of
Justice Statistics, August, 1987, pp. 6-8 and 33-34.
27
U.S. Bureau of Justice Statistics, Census of State and Federal Correctional Facilities, 1990,
NCJ-137003, Washington: U.S. Government Printing Office, May 1992, p. 18.
28
U.S. Bureau of Justice Statistics, Census of State and Federal Correctional Facilities, 1990,
NCJ-137003, Washington: U.S. Government Printing Office, May, 1992, p. 20.
29
The figures for earlier years were 10.4 percent for 1974, 6.7 percent for 1965 and 7.2 percent
for 1962; but there are real questions as to whether the differences in the pre- and post-1974
statistics reflect real change or just different data collection and reporting procedures. There is
no apparent reason for the pre- and post-1974 increase.
30
Douglas McDonald, The Price of Punishment: Public Spending for Corrections in New York,
Boulder, Colorado, Westview Press, 1980, pp. 25-30.
31
U.S. Bureau of Justice Statistics, Prisoners in State and Federal Institutions on December 31,
1983, NCJ 99861, June 1986, p. 5.
32
U.S. Bureau of Justice Statistics, Census of State and Federal Correctional Facilities, 1990,
NCJ-137003, Washington: U.S. Government Printing Office, May, 1992, p. 7.
33
John Irwin and James Austin, It’s About Time, San Francisco: National Council on Crime and
Delinquency, 1987, p. 21.
34
Christopher A. Innes, “Population Density in State Prisons,” BJS Special Report NCJ-103204,
Washington: U.S. Bureau of Justice Statistics, December, 1986, p. 7.
35
Christopher A. Innes, “Population Density in State Prisons,” BJS Special Report NCJ-103204,
Washington: U.S. Bureau of Justice Statistics, December, 1986, p. 6.
36
Patrick A. Langan, et. al., Historical Statistics on Prisoners in State and Federal Institutions,
Yearend 1925-86, Washington: U.S. Bureau of Justice Statistics, NCJ-111098, May 1988, p. 3.
37
Ibid.
38
1984 Census of State Adult Correctional Facilities, Washington: U.S. Bureau of Justice, August, 1987, pp. 28-29.
39
U.S. Bureau of the Census, Census of Governments (1987), Public Employment, No. 2, Compendium of Public Employment, Washington: U.S. Government Printing Office, p. 5.
40
Norma Mancini, Our Crowded Jails: A National Plight, Washington: U.S. Bureau of Justice
Statistics, January, 1988, pp. 2-3.
Corrections—142
41
This section was initially researched and written before the 1993 jail census results were
released. In many, but not all cases, the 1988 statistics have been updated to reflect 1993 data.
For the 1993 data, see Craig A. Perkins, James J. Stephan and Allen J. Beck, Jails and Jail
Inmates 1993-94, NCJ-151651, Washington: U.S. Bureau of Justice Statistics, April 1995, pp. 9
and 10.
42
Census of Local Jails-1988: Volume I. Selected Findings, Methodology and Summary Tables,
NCJ-127992, Washington: U.S. Government Printing Office, 1991, p. vi, and Craig A. Perkins,
James J. Stephan and Allen J. Beck, Jails and Jail Inmates 1993-94, NCJ-151651, Washington:
U.S. Bureau of Justice Statistics, April 1995, p. 5.
43
Katherine M. Jamieson and Timothy J. Flanagan, eds., Sourcebook of Criminal Justice Statistics—1988, U. S. Department of Justice, U.S. Bureau of Justice Statistics, Washington: U.S.
Government Printing Office, 1989, p. 605.
44
Craig A. Perkins, James J. Stephan and Allen J. Beck, Jails and Jail Inmates 1993-94, NCJ151651, Washington: U.S. Bureau of Justice Statistics, April 1995, p. 14.
45
Kenneth E. Kerle and Francis R Ford, The State of Our Nation’s Jails—1982, Washington:
National Sheriff’s Association, 1982, pp. 12-13 and 33.
46
Neil M. Singer and Virginia B. Wright, Cost Analysis of Correctional Standards: Institutional-Based Programs and Parole, Volume II, Washington: U.S. Law Enforcement Administration Agency, January, 1976, p. 36.
47
B. L. Wayson, et. al., The Cost of Jail Standards Compliance in Washington State, Washington: American Bar Association, 1975, p. 13.
48
U.S. Office of Management and Budget, Standard Industrial Classification Manual, Washington: U.S. Government Printing Office, 1987, pp. 409.
49
Computed from the 1990 census of prisons and the 1993 census of jails data. U.S. Bureau of
Justice Statistics, Census of State and Federal Correctional Facilities, 1990, Washington: U.S.
Government Printing Office, May 1992, and Craig A. Perkins, James J. Stephan and Allen J.
Beck, Jails and Jail Inmates 1993-94, NCJ-151651, Washington: U.S. Bureau of Justice Statistics, April 1995.
50
Ronald Goldfarb, Jails: The Ultimate Ghetto, Garden City, N.Y.: Anchor Press, 1975, p. 13.
The 1993 census of jails reported that 29 percent of the reporting institutions were unable to
provide expenditure data. See Craig A. Perkins, James J. Stephan and Allen J. Beck, Jails and
Jail Inmates 1993-94, NCJ-151651, Washington: U.S. Bureau of Justice Statistics, April 1995,
p. 9.
51
One exception is a short paper by Michael K. Block, Cost, Scale Economies and Other Economic Concepts, Washington: American Bar Association, February, 1976.
52
Michael K. Block, Cost, Scale Economies and Other Economic Concepts, Washington: American Bar Association, February, 1976, p. 34.
53
Craig A. Perkins, James J. Stephan and Allen J. Beck, Jails and Jail Inmates 1993-94, NCJ151651, Washington: U.S. Bureau of Justice Statistics, April 1995, pp. 3 and 13.
54
Ibid.
55
Kenneth E. Kerle and Francis R Ford, The State of Our Nation’s Jails — 1982, Washington:
National Sheriff’s Association, 1982, pp. 193-213, and Census of Local Jails—1983, Washington: U.S. Bureau of Justice Statistics, 1988, p. vi.
56
Census of Local Jails, 1988, NCJ 121101, Washington: U.S. Bureau of Justice Statistics,
February 1990, p. 6.
57
Census of Local Jails, 1988, NCJ 121101, Washington: U.S. Bureau of Justice Statistics,
February 1990, p. 7.
58
Craig A. Perkins, James J. Stephan and Allen J. Beck, Jails and Jail Inmates 1993-94, NCJ151651, Washington: U.S. Bureau of Justice Statistics, April 1995, pp. 6-7.
Corrections—143
59
Norma Mancini, Our Crowded Jails: A National Plight, Washington: U.S. Bureau of Justice
Statistics, January, 1988, p. 4.
60
Christopher A. Innes, Population Density in State Prisons, BJS Special Report, NCJ-103204,
Washington: U.S. Bureau of Justice Statistics, December, 1986, p. 7.
61
Christopher A. Innes, Population Density in Local Jails, 1988, BJS Special Report, NCJ122299, Washington: U.S. Bureau of Justice Statistics, March, 1990, p. 9.
62
Census of Local Jails-1988: Volume I. Selected Findings, Methodology and Summary Tables,
Washington: U.S. Government Printing Office, 1991, p. v.
63
Norma Mancini, Our Crowded Jails: A National Plight, Washington: U.S. Bureau of Justice
Statistics, January, 1988, pp. 7-8.
64
BJS Data Report, 1988, April, 1989, NCJ 116262, Washington: U.S. Bureau of Justice Statistics, p. 61.
65
Craig A. Perkins, James L. Stephan and Allan J. Beck, Jails and Jail Inmates 1993-94, Washington: U.S. Bureau of Justice Statistics, 1995, p. 1.
66
Craig A. Perkins, James L. Stephan and Allan J. Beck, Jails and Jail Inmates 1993-94, Washington: U.S. Bureau of Justice Statistics, 1995, p. 8.
67
Census of Local Jails-1988: Volume I. Selected Findings, Methodology and Summary Tables,
Washington: U.S. Government Printing Office, 1991, p. xi, 20-22.
68
Brian A. Reaves, “Sheriffs’ Departments 1990,” U.S. Bureau of Justice Statistics Bulletin,
February 1992, pp. 1-4.
69
Leslie Reiber, Parole in the United States 1980 and 1981, Washington: U.S. Bureau of Justice
Statistics, December 1985, p. 7.
70
“State and County Probation: Systems in Crisis,” Washington: U.S. General Accounting Office, May 27, 1976, p. 4.
71
Randall Guynes, “Difficult Clients, Large Caseloads Plague Probation, Parole Agencies,”
Washington: National Institute of Justice, August 1988, p. 1.
72
Ibid.
73
National Correction Reporting Program, 1990, NCJ 141879, Washington: U.S. Bureau of
Justice Statistics, May 1993, p. 23.
74
BJS Data Report, 1989, Washington: U.S. Bureau of Justice Statistics, NCJ-121514, December, 1990, p. 79; Report to the Nation on Crime and Justice, Washington: U.S. Bureau of Justice
Statistics, March, 1988, NCJ-105506, p. 105; and Probation and Parole, 1989, Washington:
U.S. Bureau of Justice Statistics, November, 1990, NCJ-125833, p. 5.
75
U.S. Office of Management and Budget, Standard Industrial Classification Manual, Washington: U.S. Government Printing Office, 1987, p. 394.
76
U.S. President’s Commission on Law Enforcement and Administration of Justice, Task Force
Report: Corrections, Washington: U.S. Government Printing Office, 1967, pp. 170-76.
77
Michigan Department of Corrections, Annual Report—1987, p. 79.
78
Douglas McDonald, The Price of Punishment: Public Spending for Corrections in New York,
Boulder, Colorado: Westview Press, 1980, p. 77.
79
Edward A. Thibault and John J. Maceri, “N.Y.S. Proactive Probation Time Management,”
Journal of Probation and Parole, 1986, pp. 17-22.
80
81
Michigan Department of Corrections, Annual Report - 1987, p. 79.
Joan Petersilia, “Probation and Felony Offenders,” Research in Brief, U.S. National Institutes
of Justice, March, 1985, p. 5.
Corrections—144
82
Alvin W. Cohn and Michael M. Ferriter, “The Presentence Investigation Report: An Old Saw
With New Teeth,” Federal Probation, September, 1990, pp. 17-18.
83
James M. Byrne, “Probation,” Crime File Study Guide, NCJ-104558, U.S. National Institutes
of Justice, 1988, p. 3.
84
Edward A. Thibault and John J. Maceri, “N.Y.S. Proactive Probation Time Management,”
Journal of Probation and Parole, 1986, pp. 17-18.
85
James M. Byrne, “Probation,” Crime File Study Guide, NCJ-104558, U.S. National Institutes
of Justice, 1988, p. 1.
86
Margaret Werner Cahalan, Historical Corrections Statistics in the United States, 1850-1984,
Washington: U.S. Bureau of Justice Statistics, December 1986, p. 7.
87
Randall Guynes, “Difficult clients, large caseloads plague probation, parole agencies,” Washington: U.S. National Institute of Justice, August 1988, p. 1-2.
88
Brian Bemus, Gary Arling, and Peter Quigley, “Workload Measures for Probation and Parole,” NIC Grant CU-4, May, 1983, p. 18.
89
Kenneth L. Hartke, “Work Units in Theory and Practice,” Corrections Today, pp. 66-68, February, 1984.
90
Joan Petersilia and Susan Turner, “Comparing Intensive and Regular Supervision for High
Risk Probationers,” Crime and Delinquency, January, 1990, p. 94.
91
Correctional Populations in the United States 1985, NCJ-103957, Washington: U.S. Bureau
of Justice Statistics, December 1987, pp. 27 and 92; and Tracy L. Snell, Correctional Populations in the United States, 1992, NCJ-146413, Washington: U.S. Bureau of Justice Statistics,
January 1995, pp. 27 and 106.
92
Douglas McDonald, The Price of Punishment: Public Spending for Corrections in New York,
Boulder, Colorado: Westview Press, 1980, p. 73.
93
Douglas McDonald, The Price of Punishment: Public Spending for Corrections in New York,
Boulder, Colorado: Westview Press, 1980, p. 148.
94
“Arizona Criminal Justice System Research Projects, Sub-project II, Arizona—Study of the
Cost of Incarceration and Rehabilitation.” Study performed by Lawrence-Leiter and Company,
Kansas City, Missouri, p. 121. (No publication date.)
95
Kathleen Maguire and Ann L. Pastore (eds.), Sourcebook of Criminal Justice Statistics —
1993, Washington: U.S. Government Printing Office, 1994, p. 43.
96
Justice Expenditure and Employment in the United States, 1988, NCJ-125619, Washington:
U.S. Government Printing Office, 1990(?), pp. 118-19.
97
Kathleen Maguire and Ann L. Pastore, eds., Sourcebook of Criminal Justice Statistics 1993.
U.S. Bureau of Justice Statistics, Washington: Government Printing Office, 1994, p. 585. This
section was researched and written prior to the collection of the 1993 juvenile survey data. In
many, but not all, cases the data have been updated to reflect the 1993 statistics.
98
Personal communication from Office of Juvenile Justice and Delinquency Prevention, U.S.
Department of Justice, January 25, 1993.
99
Kathleen Maguire and Ann L. Pastore, eds., Sourcebook of Criminal Justice Statistics 1993.
U.S. Bureau of Justice Statistics, Washington: U.S. Government Printing Office, 1994, p. 585.
100
Children in Custody, 1973. No SD-JD-2F, U.S. Department of Justice, December 1977, pp. 6
and 13; and personal communication from Office of Juvenile Justice and Delinquency Prevention, January 25, 1993.
101
Kathleen Maguire and Ann L. Pastore, eds., Sourcebook of Criminal Justice Statistics 1993.
U.S. Bureau of Justice Statistics, Washington: U.S. Government Printing Office, 1994, p. 12,
and Justice Expenditure and Employment: 1988, NCJ-125619, U.S. Bureau of Justice Statistics,
Washington: U.S. Government Printing Office, 1991, pp. 136-7.
Corrections—145
102
Kathleen Maguire and Ann L. Pastore, eds., Sourcebook of Criminal Justice Statistics 1993.
U.S. Bureau of Justice Statistics, Washington: U.S. Government Printing Office, 1994, pp. 581
and 585-6.
103
Kathleen Maguire and Ann L. Pastore, eds., Sourcebook of Criminal Justice Statistics 1993.
U.S. Bureau of Justice Statistics, Washington: U.S. Government Printing Office, 1994, pp. 584
and 586.
104
Children in Custody, 1975-85, NCJ-114065, Washington: U.S. Bureau of Justice Statistics,
May 1989, pp. 3, 25-27 and 32.
105
“Juvenile Detention Using Staff Supervision Rather than Architectural Barriers,” Washington: U.S. General Accounting Office, December, 1985, pp. 1-2.
106
Juvenile Justice and Delinquency Prevention Act of 1974 as amended (42 USC 5601).
107
Children in Custody, 1975-85, NCJ-114065, Washington: U.S. Bureau of Justice Statistics,
May 1989, pp. 37 and 39, and Margaret Werner Cahalan, Historical Correction Statistics in the
United States, 1850-1984, Washington: U.S. Bureau of Justice Statistics, NCJ-102529, December, 1986, pp. 120 and 124.
108
Children in Custody, 1975-85, NCJ-114065, Washington: U.S. Bureau of Justice Statistics,
May 1989, p. 39; and personal communication from Office of Juvenile Justice and Delinquency
Prevention, U.S. Department of Justice, December 1992.
109
“Juveniles Processed in Criminal Court and Case Dispositions,” Washington: U.S General
Accounting Office, August, 1995, p. 4.
110
U.S. Bureau of Justice Statistics, Report to the Nation on Crime & Justice, Washington: U.S.
Government Printing Office, March 1988, p. 79.
111
U.S. Office of Management and Budget, Standard Industrial Classification Manual, Washington: U.S. Government Printing Office, 1987, pp. 395-6.
112
Employment and Earnings, January 1993, p. 96.
113
U.S. Bureau of the Census, Public Employment in 1971, Series GE-No. 1, Washington: U.S.
Government Printing Office, 1972, p. 9, and U.S. Bureau of the Census, Public Employment:
1991, Series GE/91-1, Washington: U.S. Government Printing Office, 1992, p. 11.
114
Margaret Werner Cahalan, Historical Corrections Statistics in the United States, 1850-1984,
Washington: U.S. Bureau of Justice Statistics, NCJ-102529, December 1986, pp. 101-103.
115
Selected issues of Children in Custody and private communications with Office of Juvenile
Justice and Delinquency Prevention, U.S. Department of Justice.
116
Children in Custody; Public Juvenile Facilities, 1987, Washington: U.S. Office of Juvenile
Justice and Delinquency Prevention, October, 1989, p. 2.
117
John Irwin and James Austin, Its About Time, San Francisco: National Council on Crime and
Delinquency, 1987, p. 21.
118
Norma Mancini, Our Crowded Jails: A National Plight, Washington: U.S. Bureau of Justice
Statistics, NCJ-111846, June 1988, pp. 4-5; and Christopher A. Innes, Population Density in
State Prisons, Washington: U.S. Bureau of Justice Statistics, Special Report NCJ-103204, December, 1986, pp. 6-7.
119
Personal communication, Office of Juvenile Justice and Delinquency Prevention, U.S. Department of Justice, December 15, 1992.
120
121
Ibid.
Children in Custody—1975, No SD-JD-4F, Washington: U.S. Law Enforcement Assistance
Administration, December, 1979, p. 75.
Corrections—146
122
Non-payroll staff employees are individuals who work in juvenile facilities, but who are paid
by other governmental and non-governmental organizations such as education and health.
123
Personal communication, Office of Juvenile Justice and Delinquency Prevention, U.S. Department of Justice, January 25, 1993.
124
U.S. Bureau of the Census, Public Employment: 1992, Series GE/92-1, Washington: U.S.
Government Printing Office, 1994, p. 5.
125
Ibid.
126
Census of Local Jails-1988, Washington: U.S. Bureau of Justice Statistics, February, 1990,
p. 8.
Corrections—147
Chapter 5. State Employment Security
Services
T
his chapter discusses labor productivity measurement of two, long-standing State
government employment security services, the Unemployment Insurance pro
gram and the Employment Service. The Unemployment Insurance (UI) program provides stipends for eligible unemployed workers. The Employment Service
(ES) helps individuals obtain jobs. The Federal Government provides grants for administration of the programs and oversees their operations.
Although the two services are often operated out of the same office, they have very
different goals, missions, objectives, and management. The UI is an income maintenance program supported largely by a tax on employers. UI State personnel screen
applicants, certify their eligibility for assistance, and write checks. Most UI funds go to
pay the unemployed. The Employment Service, on the other hand, is a job assistance
program. Its staff tests client job skills, prepares job listings, and refers clients to job
openings or training.
In 1992, these two services spent about $41 billion. UI expenditures were about $40
billion of which 94 percent was for unemployment benefits and the balance for administrative expenses. The $1 billion spent by ES covered program and administrative
expenditures. Despite their long history, the UI and ES are relatively modest programs
in terms of the number of employees, probably totaling fewer than 75,000 State employees in 1992.
These two programs are noted for their measurable outputs, available Federal data,
past research, and Federal funding and interest. State government employees operate
these joint Federal-State programs.
The first part of this chapter discusses the Unemployment Insurance program; this is
followed by a discussion of the Employment Service. In each case, the program is
briefly described, potential output measures are examined, data to calculate the indexes
are presented, and indexes are calculated. The final section of the chapter combines the
two services into a single employment security index.
Unemployment
Insurance
Unemployment Insurance is a joint Federal-State program that has existed for over
half a century. It has a specified mission, a clear cut set of outputs, and good national
data. A number of studies have examined the program, and BLS published a national
productivity index for several years. The following discussion reviews, updates, and
summarizes some of the indexes that have been published for the UI.1
Institutional setting
The Social Security Act of 1935 established the UI program. The program’s primary
aim is to aid persons temporarily unemployed through no fault of their own. Another
objective is to increase aggregate demand during times of rising unemployment to help
stabilize the economy.2 Federal laws and regulations set broad operational guidelines;
State laws, regulations, and procedures govern day-to-day operations. Every State and
some territories operate a UI program.
UI coverage is provided through several different programs (table 65). Three—the
regular State program for unemployed workers, the program for unemployed ex-service members, and the program for unemployed Federal workers—comprise the bulk
of the workload throughout the period examined here, 1964-92. In addition, several
national programs operated for part of the period, including the extended benefit program, the temporary compensation program, the Federal supplemental benefit program,
State Employment Security Services—148
and the Emergency Unemployment Compensation Program. Data from these programs
are also included. Several special programs, including the Trade Adjustment Act and
the Redwood Employees Protection program, also operated during this period, produce
different outputs, and are too small to be included here. The regular State program is
the backbone of the UI and normally accounts for most of the benefit expenditures.
Table 65. Selected Unemployment Insurance programs, 1935-92.
Program or law
Period covered
1.
Regular State programs (UI-REG)/PL 74-271
1935 to present
2.
Unemployment Compensation for Federal Employees
(UCFE)/PL 83-767
1954 to present
Unemployment Compensation for Ex-servicemen
(UCX)/PL 85-848
1958 to present
4.
Extended Benefits (EB)/PL 91-373
1970 to present
5.
Emergency Unemployment Compensation Act Temporary
Compensation (TC)/PL 92-224
1971 - 1973
6.
Federal Supplemental Benefits (FSB)/PL 93-572
1975 - 1978
7.
Special Unemployment Assistance (SUA)/PL 93-576
1974 - 1978
8.
Disaster Unemployment Assistance (DUA)/PL 93-288
1974 to present
9.
Trade Adjustment Assistance (TAA)/PL 93-610
1974 to present
3.
10. Federal Supplemental Compensation (FSC)/PL 97-248
1982 - 1985
11. Emergency Unemployment Compensation/PL 102-164
1991 - 1994
SOURCE: Paul L. Burgess and Jerry L. Kingston, An Incentives Approach to Improving the Unemployment Compensation System, Kalamazoo, Michigan: W.E. Upjohn Institute for Employment Research, 1987,
pp. 82-4, and Saul J. Blaustein, Unemployment Insurance in the United States, The First Half Century,
Kalamazoo, Michigan: W.E. Upjohn Institute for Employment Research, 1993.
The Unemployment Insurance program has two basic missions. One focuses on the
unemployed, that is, screening applicants and paying those who qualify for support.
The other focuses on funding or finances, that is, collecting taxes from employers to
finance the program. In fiscal 1992, service to the unemployed, or beneficiary services,
accounted for about 67 percent of the Unemployment Insurance program labor input;
the funding or financial mission accounted for about 20 percent; and support and overhead, made up the remaining 13 percent.
Beneficiary services, or services to the unemployed, include screening applicants,
determining their eligibility, hearing appeals, calculating benefit payments, and issuing
checks. In FY 1992, 28.3 million claims were filed, and 223.4 million weeks of compensation were paid.
Applicant eligibility requirements are set by each State and include such considerations as the reason for leaving the job, wages required to qualify for unemployment
insurance coverage, earnings from part-time work when drawing unemployment insurance, and the length of time worked to qualify. Also, each State sets its own rules for the
weekly amount paid to recipients.
Financial services or tax collection, the other Unemployment Insurance mission, is a
function of the number and type of employers. State governments collect taxes from
employers to pay benefits to the unemployed, and the Federal Government, through the
Internal Revenue Service, collects taxes from employers to pay administrative costs.
State Employment Security Services—149
General Federal revenue pays benefits to unemployed veterans and Federal workers,
and for some special programs, such as Emergency Unemployment Compensation and
Supplementary Unemployment Assistance. Other special programs have been funded
jointly by the Federal and State governments. Financial operations collect money from
employers to support the UI payments, audit employers’ records, and track down delinquent accounts. Most employers and employees are covered by UI operations. In
1992, 5.7 million employers were taxed to support UI programs. About 105 million
individuals or 92 percent of the U.S. civilian labor force was covered. 3
Unemployment Insurance costs exceeded $40 billion in 1992. Benefits were about
$37.5 billion, and State administrative costs were about $2.5 billion. In addition, there
was about $.3 billion in assorted Federal expenditures.4 About 46,600 State staff years
were devoted to UI operations in 1992.
Neither the Census of Governments nor the Standard Industrial Classification (SIC)
system separately identify the UI program. The Census of Governments includes the
UI program under the general category of Social Insurance Administration, while the
SIC assigns the program to Industry 9441, Administration of Social, Manpower and
Income Maintenance Programs.5 That is, the Standard Industrial Classification Manual
lumps the program with public welfare administration, equal employment opportunity
offices, medical assistance program administration, and worker’s compensation offices.
The data used to compute the UI indexes are taken primarily from those that are
routinely collected by the U.S. Department of Labor.
Outputs
Selection of a measure to track Unemployment Insurance outputs is relatively straightforward, at least when compared with other government services. Although no research apparently addresses the issue, there are several obvious candidate measures:
Number of—
• Employees covered
• Employers covered
• Beneficiaries
• Compensated weeks
• Claims processed
In addition to these five measures, three others are also candidates for consideration:
• Benefit index
• Finance index
• Composite program index
Each measure is defined, its strengths and weaknesses are listed, and when appropriate, an index is presented for the measure. Chapter 2 presents the criteria used to
evaluate each measure.
Number of employees
covered
Number of employers
covered
The number of employees measure is a count of persons who have earned wages in
jobs covered by unemployment insurance. A similar measure is often used by the private insurance industry where “the number of policies sold” is a measure of output. The
arguments for using this output measure are the following: (1) It is measurable; (2) it
has been calculated for a number of years; (3) it is a physical measure; (4) it is easily
understood; and (5) it is supported by good data.
The primary argument against using this measure is that it is not the final product of
the UI. There is little connection between the number of persons covered and the resources required to operate the UI. Hence, this measure is not considered further in this
discussion.
The number of employers measure is a count of the businesses, firms, and organizations, which have one or more individuals who are covered by unemployment insurance. General Motors Corp. and Montgomery County, MD, are examples of covered
State Employment Security Services—150
employers. The arguments for using the number of employers as the UI output measure
are similar to those noted in the preceding discussion. The primary argument against it
as a measure is that there is little correlation between the number of employers and
overall UI operations.
However, there is one part of the UI operation, the finance activity, where there is a
close relationship between the number of employers and the work performed by the
staff. The finance activity monitors employer’s contributions, audits employer’s records
to ensure compliance with Federal and State laws, and attempts to collect monies owed
on delinquent accounts. This part of the UI index, which is presented later, uses the
number of employers as the basis for its calculations.
Number of beneficiaries
The number of beneficiaries is the number of individuals who draw unemployment
insurance benefits. There are several variations of this measure: The total number of
different individuals (one person may draw unemployment benefits several times during a year); the number of different times a person is assisted during the course of a
year; and the number of claims. Claimants and beneficiaries, though closely related,
are not synonymous; individuals can file claims but do not become beneficiaries until
the claim is approved.
The number of beneficiaries is measurable, repetitive, and easily understood and
data are available with which to calculate an index. This measure is sometimes used as
an output measure of the UI program.
However, the measure fails to take into account the length of time an individual
draws unemployment insurance. During periods of high unemployment, the average
length of unemployment increases and the number of tasks (check writing, recertifications, and appeals) associated with the maintenance of a person collecting unemployment compensation also increases. These tasks require additional labor, which is not
accounted for in this measure. Because of this weakness a different measure is used in
the calculations.
Number of compensated
weeks
Number of claims
processed
A measure which takes into account the length of time unemployment insurance is
drawn, as well as the number of people drawing insurance, is the number of compensated or beneficiary weeks. This measure is the aggregate number of weeks for which
unemployment insurance is paid during a given period, such as a year.
One variation of the number of compensated weeks is the “average weekly number
of beneficiaries” for a year, which is simply the total divided by 52. Another is the
“average weekly insured unemployed (AWIU),” which includes all eligible persons
claiming at least 1 week of unemployment insurance during the reporting period.
The number of compensated weeks is sometimes used as the preferred output measure for unemployment insurance. It is measurable, repetitive, accurate, and comparable through time and different State unemployment insurance levels don’t affect it.
Also, it is an easily understood, unitary measure and data exist which can be used to
calculate an output index for a number of years.
The primary argument against this measure is that compensation is only part of the
UI program. Other parts are financing and application. However, the number of compensated weeks paid does reflect an important part of UI staff work, and it is included as
part of the program index presented below.
The count of the number of compensated weeks is extremely susceptible to the time
period examined. In 1964, there were 79.8 million compensated weeks paid; in 1992
the figure had increased to 223.4 million. But, the annual statistic has been as low as
49.2 million in 1969 and as high as 260 million in 1976. The average annual increase
between 1964 and 1992 is 3.7 percent (table 66).
Claim processing is the activity whereby applications for unemployment insurance
are taken by the UI staff. Not every claim leads to a UI payment, but most do. Hence,
processing of a claim is a necessary but not a sufficient condition to issue a UI check.
Claim processing is an activity that needs to be included in any calculation of UI output,
State Employment Security Services—151
but it is not an adequate measure of total UI output in and of itself.
As discussed earlier, there were several different types of UI programs during the
covered period: State, Federal, veteran’s, special unemployment assistance, Federal
supplementary compensation, and Federal supplementary benefits. Application processing time varies by UI program. Applications for the regular UI program take about
half the time to process as a Federal claim.
In addition, if a beneficiary experiences a second term of unemployment during the
same benefit year, an additional claim must be filed. Because much of the required
information is already on file and has been verified, the time required to process an
additional claim is about one-quarter of the initial claim. In some years, additional
claims are almost one-third of total claims.
There were 15.3 million initial claims filed in 1964 and 18.6 million in 1992, but in
1975, 30.3 million were filed. There were only 10.5 million initial claims in 1969. The
number of additional claims has ranged from a high of 13.6 million in 1982 to a low of
6.1 million in 1989.
Because of the number of different types of claims and the difference in the time to
process each type, a labor weighted claims index has been prepared to reflect claims
processing operations. That is, initial and additional claims for UI programs are separately weighted by their processing time before being combined into a single index.
The results of these computations show that the average annual increase in the weighted
index is 1.7 percent for 1964-92 (table 66). The unweighted figure is 2.2 percent.
Benefit index
A good case can be made for dividing UI outputs into two parts: service to the
unemployed, that is, those applying for and drawing benefits and finance, or tax collection activities, that is, those activities that relate to the financing of UI operations. This
section discusses the benefit index.
Service to the unemployed is a function of the number applying for and the number
drawing UI benefits. The number of claims processed weighted by program and type of
claim is used to measure the number of claims. Benefits are a measure of the number of
weeks compensated. These output indexes are combined using staff year labor weights
to produce the benefit index.
The average annual increase in the benefit index between 1964 and 1992 is 2.7
percent (table 66).
Finance index
The other part of UI operations is the collection of funds to pay beneficiaries and
operate the program. As noted above, this part of UI operations monitors employers’
contributions, audits employer’s records, and attempts to collect monies owed on delinquent accounts. One approach to measuring finance operations is to measure each of
the activities and weight and combine them to create the finance index. This is a fairly
laborious task given the nature of the work, and it is questionable as to whether the data
are available with which to make such calculations.
Hence, the approach used here is to count the number of employers. Because there
is a close relationship between the number of employers and the work performed by the
UI finance or employer activity staff, this is a satisfactory proxy measure. Counts of the
number of employers show that there has been a continuous increase in the number over
the period covered by the index. In 1964 there were 2.4 million employers and by 1992
the figure had reached 5.7 million. The average annual increase was 3.2 percent, and
every year shows an increase (table 66).
Program index
To measure total Unemployment Insurance program output, the output associated
with service to the unemployed is combined with the output associated with finance
operations. The procedure uses staff year labor weights to produce a weighted index of
output for the entire Unemployment Insurance program. The results of this calculation
show that output grew by 2.5 percent annually between 1964 and 1992 (table 66).
State Employment Security Services—152
Comparison of output
indexes
Eight measures of UI output have been discussed in this section. Indexes have been
calculated for five of them: Claims processed, weeks compensated, benefits, finance,
and total program. Comparison of the trends for the five indexes for 1964-92 show that
weeks compensated grew most rapidly (3.7 percent), followed by finance (3.2 percent),
benefit (2.7 percent), program (2.5 percent), and claims (1.7 percent). The benefit and
program indexes show similar rates of growth because such a large part of the program
index reflects the benefit index. In 1992, 67 percent of the direct labor was used to
provide benefit payments.
The rate of growth of each index depends largely on the period examined. Rates
have been calculated for seven periods. They are:
•
•
•
1964-92, the entire measured period,
1967-92, the period covered by the productivity indexes presented for most of
the other indexes in this study, and
1967-72, 1972-76, 1976-80, 1980-83 and 1983-92, the peak periods for the
measured program output.
Peak work periods for the UI program generally coincide with economic recessions.
The business cycle troughs for the total economy for 1964-92 were 1970, 1975, 1980,
1982, and 1991 as determined by the National Bureau of Economic Research.
Several points stand out in these indexes. First, the period chosen can dramatically
affect the rate of growth. With the exception of the finance index, the rate of change
seems to depend as much on the time period chosen as on the index chosen. This is a
reflection of the business cycle. Second, individual activity indexes often peak in different years. Although the benefit and program indexes peak in the same years, this is
not the case for claims and weeks compensated. The peaks in claims processed usually
lead the peaks in weeks compensated by about 1 year. Third, there is greater variability
in weeks compensated than in claims processed. Fourth, there is very little difference
between the rate of change in the benefit and program indexes, regardless of the period.
Fifth, for most years, the program index falls between the benefit and finance indexes as
expected. However, in 4 years, 1980-82 and 1992, the program index fell below the
two other indexes. This counter-intuitive result reflects the large increase in the weeks
compensated index, and the change in base year labor weights, which increased the
relative influence of the claims processed index.
Quality of UI output
UI program quality output is often studied and discussed. It is an area that is thoroughly documented, and the national UI office has collected statistics on UI quality for
several decades. Today, statistics are collected on several dozen quality variables, which
are summarized and reported annually.6 Although there is criticism of the UI quality
measurement system, it is one of the few government programs that attempt to measure
the quality of its output in a systematic manner.7
The question for this study of productivity is: how does quality affect UI output and,
particularly, how should quality change be reflected in output trends? To affect output
trends, two conditions must exist: First, output quality must be changing and second,
the quality change must affect resource inputs. Some quality characteristics, such as
courtesy and helpfulness of UI staff, are important to UI managers but probably do not
affect resource requirements and thus do not directly concern this investigation. Other
characteristics, such as error rates, could affect resource requirements but apparently
are fairly stable through time.
A cursory review of UI quality measurements does not find an overall shift in UI
output quality during the period covered here. Some measures show improvement,
some deterioration, but overall there is little change. The output and productivity indexes presented here have not been adjusted to account for quality shifts.
State Employment Security Services—153
Table 66. Comparison of five Unemployment Insurance output indexes, 1964-92.
(1964 = 100)
Fiscal
year
Total
program
Claims
processed
Weeks
compensated
Benefit
Finance
1964 ...................
1965 ...................
1966 ...................
1967 ...................
1968 ...................
1969 ...................
100.0
90.1
79.7
81.1
80.7
77.8
100.0
87.3
74.5
76.8
74.0
68.4
100.0
84.1
66.3
64.9
66.8
61.7
100.0
85.8
70.5
71.0
70.3
64.9
100.0
100.9
102.4
106.2
106.2
107.6
1970 ...................
1971 ...................
1972 ...................
1973 ...................
1974 ...................
1975 ...................
1976 ...................
1977 ...................
1978 ...................
1979 ...................
90.8
115.6
119.5
108.1
116.4
193.7
219.5
184.9
146.0
136.3
87.5
111.3
110.6
89.0
102.4
191.8
183.3
167.3
131.9
124.8
80.3
137.5
146.2
110.8
115.7
241.9
325.9
237.1
161.1
136.8
83.7
122.1
125.7
97.9
107.2
212.5
246.9
197.4
144.4
129.7
111.1
112.4
117.2
153.2
159.8
163.8
167.2
172.9
183.0
191.6
1980 ...................
1981 ...................
1982 ...................
1983 ...................
1984 ...................
1985 ...................
1986 ...................
1987 ...................
1988 ...................
1989 ...................
171.8
167.5
198.2
218.4
153.6
153.0
148.1
140.0
133.2
133.5
159.3
144.0
182.7
177.8
129.9
131.7
123.9
111.9
104.1
104.9
194.9
200.7
240.2
301.9
171.8
162.7
155.6
142.8
122.3
120.8
174.5
168.5
207.4
233.5
147.9
144.7
137.3
125.0
111.8
111.6
197.0
198.7
201.9
203.4
206.8
216.4
219.9
226.8
232.6
234.3
1990 ...................
1991 ...................
1992 ...................
144.5
168.4
199.2
117.8
143.4
160.6
139.9
192.0
280.0
127.1
164.1
211.6
238.4
239.9
242.1
Average annual
percent change:
1964-92 ..............
1967-92 ..............
2.5
3.7
1.7
3.0
3.7
6.0
2.7
4.5
3.2
3.4
1967-72 ..............
1972-76 ..............
2.2
16.4
1.3
13.5
4.9
22.2
2.9
18.4
2.0
9.3
1976-80 ..............
1980-83 ..............
1983-92 ..............
-5.9
8.3
-1.0
-3.4
3.7
-1.1
-12.1
15.7
-.8
-8.3
10.2
-1.2
3.3
1.1
2.0
Labor inputs
UI program operations are very labor intensive. About 80 percent of all administrative funds go to pay employee salaries and benefits. Building rents, computer leases,
telephone, postage, and the like account for the remaining 20 percent. All operating
personnel are State government employees. In addition to the State employees, there
are about 150 Federal Government employees who oversee and coordinate State activities, but they are not included in the labor or cost figures presented in this bulletin.
Two labor measures are recommended for calculating State and local government
labor productivity—total employment and the number of full-time-equivalent employ-
State Employment Security Services—154
ees. The data used in this section for computational purposes are the actual number of
UI State filled positions (taken from accounting reports). State labor officials report
these data to the U.S. Department of Labor. The number of positions is equivalent to an
FTE, and is used to budget for UI programs and to account for funds allocated to the
States. The trend computed from the number of State filled positions would approximate an hours trend if it were possible to compute that index. However, data are not
available with which to calculate an index of the total number of full-time and part-time
UI employees.
The total number of UI positions increased from 32,212 in 1964 to 46,465 in 1992.
The average annual rate of increase between 1964 and 1992 is 1.3 percent, with major
fluctuations following workload shifts (table 67). The number of staff years ranged
from a low of 26,070 in 1967 to a high of 57,208 in 1976.8
The labor statistics presented here support the basic unemployment insurance program including the regular State program, Federal, Ex-service Members, Extended
Benefits, Supplementary Benefits, Special Unemployment, and Temporary Compensation (table 67). Positions for trade adjustment assistance, disaster unemployment assistance, and redwood assistance have been removed from the totals because outputs for
these programs are not included in the program outputs presented in this bulletin.
Table 67. Number and index of State UI labor positions, 1964-92.
(1964 = 100)
Fiscal year
Number
Index
1964 .............................................
1965 .............................................
1966 .............................................
1967 .............................................
1968 .............................................
1969 .............................................
32,212
31,543
27,602
26,070
27,334
27,016
100.0
97.9
85.7
80.9
84.9
83.9
1970 .............................................
1971 .............................................
1972 .............................................
1973 .............................................
1974 .............................................
1975 .............................................
1976 .............................................
1977 .............................................
1978 .............................................
1979 .............................................
28,489
31,884
37,799
34,581
32,711
44,528
57,208
55,125
49,070
48,684
88.4
99.0
117.3
107.4
101.5
138.2
177.6
171.1
152.2
151.1
1980 .............................................
1981 .............................................
1982 .............................................
1983 .............................................
1984 .............................................
1985 .............................................
1986 .............................................
1987 .............................................
1988 .............................................
1989 .............................................
48,673
51,350
52,907
56,351
46,512
44,652
41,336
39,778
38,581
38,527
151.1
159.4
164.2
174.9
144.4
138.6
128.3
123.5
119.8
119.6
1990 .............................................
1991 .............................................
1992 .............................................
37,849
40,764
46,465
117.5
126.5
144.2
Average annual
percent change:
1964-1992 ....................................
1967-1992 ....................................
—
—
1.3
2.3
NOTE: Excludes disaster, trade adjustment, and redwood program personnel.
State Employment Security Services—155
Position data are available by function for the UI system for a number of years. The
UI program has a base staff year level that does not vary significantly nationally. The
contingency staff year level is adjusted in accordance with the workload. Major fluctuations in personnel can follow shifts in workload.
Labor indexes were also calculated for the individual output functions discussed in
the preceding section, that is, claims processed, weeks compensated, benefits (claims
and compensation), and finance (table 68). In each instance, the labor indexes reflect
only the direct labor used to produce the functional output indexes (table 66). Support
staff are not assigned by function, and it is not possible to make such an assignment in
any reasonable fashion. Because the direct-indirect labor ratio varies through time, the
rate of change and the total amount of labor covered by these functional indexes is
different from the overall UI labor index presented in table 67.
Table 68. Four UI direct labor indexes, 1964-92.
(1964 = 100)
Fiscal
year
Claims
processed
Weeks
compensated
Benefit
Finance
1964 ..................
1965 ..................
1966 ..................
1967 ..................
1968 ..................
1969 ..................
100.0
94.3
81.7
77.6
79.5
74.9
100.0
90.0
77.7
71.9
74.1
71.1
100.0
91.2
78.9
73.5
75.7
72.1
100.0
100.3
102.4
100.1
99.4
99.7
1970 ..................
1971 ..................
1972 ..................
1973 ..................
1974 ..................
1975 ..................
1976 ..................
1977 ..................
1978 ..................
1979 ..................
86.2
130.3
142.5
112.8
117.8
251.6
278.7
240.2
187.5
158.0
74.3
108.3
122.1
102.0
96.1
156.5
199.3
187.0
162.2
144.2
77.7
114.5
127.9
105.1
102.3
183.5
221.9
202.1
169.4
148.1
102.2
96.6
107.5
116.8
112.3
111.2
119.8
124.7
124.4
126.3
1980 ..................
1981 ..................
1982 ..................
1983 ..................
1984 ..................
1985 ..................
1986 ..................
1987 ..................
1988 ..................
1989 ..................
197.7
182.2
217.1
219.1
149.0
160.1
138.2
142.6
124.5
128.2
179.8
171.3
191.8
190.6
129.1
136.8
120.0
115.6
109.1
108.6
184.9
174.6
198.8
198.4
134.5
143.1
125.0
122.7
113.3
113.9
137.0
138.5
140.4
134.4
133.3
134.0
137.0
142.8
148.2
148.3
1990 ..................
1991 ..................
1992 ..................
124.4
143.6
179.1
105.4
118.6
146.7
110.5
125.3
155.3
150.8
154.0
153.8
Average annual
percent change:
1964-92 .............
1967-92 .............
2.1
3.4
1.4
2.9
1.6
3.0
1.6
1.7
State Employment Security Services—156
The fact that the benefit and finance labor indexes increased at the same average
annual rate of 1.6 percent between 1964-92 is largely happenstance; the 1967-92 years
yield very different growth rates. Examination of the individual functional indexes reveal that the finance labor index increased at a fairly steady rate between 1964-92. The
benefit labor index, on the other hand, registered wide fluctuations reflecting the fluctuations in workload in claim processing and weeks compensated. The long-term average annual increase in claims processed is 2.1 percent and weeks compensated is 1.4
percent. But it is the large number of year-to-year fluctuations that is the most interesting aspect of these indexes (table 68).
Productivity trends
UI program output per employee increased 1.2 percent annually between 1964-92,
and 1.3 percent between 1967-92. The rates of growth are directly affected by changes
in the economy, and, in particular, by the changes in insured unemployment. Most interesting are the large year-to-year fluctuations and cyclical movements. From 1967
through 1992, there were changes in output per employee year of 10 percent or more in
12 of the 25 measured years (table 69).
Examination of the five output cycles between 1967-92 (1967-72, 1972-76, 197680, 1980-83, and 1983-92) shows that output per employee year generally rose as output increased, although obviously not at the same rate. Although staff was added, they
were not added as rapidly as output increased. The result is increasing output per employee year. The opposite seems to have happened as output dropped. What is most
noteworthy about this process is that, in most cases, staff were added during the upswing of the output cycles and cutback on the downside, a phenomenon common in the
private sector, but not usually found in public operations.
Output per employee year was also calculated for the four UI functions—benefits,
claims, weeks compensated and finance—for which we calculated output and labor
indexes. The labor indexes, as noted earlier, reflect direct labor only. Computations of
output per employee year show that the benefit index increased at about the same rate
as the overall UI index (1.4 versus 1.3 percent per year for 1967-92). However, the two
parts of the benefit index, claims processed and weeks compensated, show very different movements. The claims output per employee year index declined (-0.4) over the
1967-92 period, and showed relatively minor year-to-year fluctuations. The weeks
compensated index, on the other hand, showed moderate rates of growth (3.0) and
considerable fluctuations from year to year. Finally, the finance index shows a modest
rate of growth with an average annual increase between 1967-92 of 1.6 percent. However, even this index shows two distinct periods of change. From 1967 through 1975,
output per employee year increased at an average annual rate of 4.2 percent, but since
then the average annual increase was only 0.4 percent (table 70).
State Employment Security Services—157
Table 69. Unemployment Insurance productivity index,
1964-92.
(1964 =100)
Output per
employee
Fiscal
year
Output
Employee
positions
1964 ........................
1965 ........................
1966 ........................
1967 ........................
1968 ........................
1969 ........................
100.0
90.1
79.7
81.1
80.7
77.8
100.0
97.9
85.7
80.9
84.9
83.9
100.0
92.0
93.0
100.2
95.1
92.8
1970 ........................
1971 ........................
1972 ........................
1973 ........................
1974 ........................
1975 ........................
1976 ........................
1977 ........................
1978 ........................
1979 ........................
90.8
115.6
119.5
108.1
116.4
193.7
219.5
184.9
146.0
136.3
88.4
99.0
117.3
107.4
101.5
138.2
177.6
171.1
152.3
151.1
102.7
116.8
101.8
100.7
114.6
140.2
123.6
108.1
95.8
90.2
1980 ........................
1981 ........................
1982 ........................
1983 ........................
1984 ........................
1985 ........................
1986 ........................
1987 ........................
1988 ........................
1989 ........................
171.8
167.5
198.2
218.4
153.6
153.0
148.1
140.0
133.2
133.5
151.1
159.4
164.2
174.9
144.4
138.6
128.3
123.5
119.8
119.6
113.7
105.1
120.7
124.8
106.4
110.4
115.4
113.4
111.2
111.6
1990 ........................
1991 ........................
1992 ........................
144.5
168.4
199.2
117.5
126.5
144.2
123.0
133.0
138.1
Average annual
percent change:
1964-92 ...................
1967-92 ...................
2.5
3.7
1.3
2.3
1.2
1.3
1967-72 ...................
1972-76 ...................
1976-80 ...................
1980-83 ...................
1983-92 ...................
8.1
16.4
-5.9
8.3
-1.0
7.7
10.9
-4.0
5.0
-2.1
.3
5.0
-2.1
3.2
1.1
State Employment Security Services—158
Table 70. Comparison of four UI direct labor productivity indexes,
1964-92
(1964 = 100)
Fiscal
year
Claims
processed
Weeks
compensated
Benefit
Finance
1964 ...................
1965 ...................
1966 ...................
1967 ...................
1968 ...................
1969 ...................
100.0
92.6
91.2
98.9
93.1
91.4
100.0
93.5
85.3
90.3
90.1
86.7
100.0
94.0
89.4
96.5
92.9
90.0
100.0
100.5
100.0
106.0
106.8
108.0
1970 ...................
1971 ...................
1972 ...................
1973 ...................
1974 ...................
1975 ...................
1976 ...................
1977 ...................
1978 ...................
1979 ...................
101.5
85.4
77.6
78.9
86.9
76.2
65.8
69.6
70.4
79.0
108.1
127.0
119.8
108.6
120.4
154.6
163.5
126.8
99.3
94.8
107.7
106.6
98.3
93.2
104.9
115.8
111.3
97.6
85.2
87.6
108.7
116.3
109.0
131.2
142.2
147.4
139.6
138.7
147.1
151.7
1980 ...................
1981 ...................
1982 ...................
1983 ...................
1984 ...................
1985 ...................
1986 ...................
1987 ...................
1988 ...................
1989 ...................
80.6
79.1
84.2
81.2
87.1
82.3
89.7
78.4
83.6
81.8
108.4
117.2
125.2
158.3
133.1
118.9
129.7
123.5
112.2
111.2
94.4
96.5
104.4
117.7
110.0
101.1
109.9
101.9
98.7
98.0
143.7
143.4
143.8
151.4
155.1
161.5
160.6
158.9
156.9
158.0
1990 ...................
1991 ...................
1992 ...................
94.7
99.8
89.7
132.7
161.9
190.9
115.1
131.0
136.3
158.1
155.7
157.5
Average annual
percent change:
1964-92 ..............
1967-92 ..............
-0.4
-.4
2.3
3.0
1.1
1.4
1.6
1.6
1967-72 ..............
1972-76 ..............
1976-80 ..............
1980-83 ..............
1983-92 ..............
-4.7
-4.0
5.2
.2
1.1
5.8
8.1
-9.8
13.4
2.1
.4
3.2
-4.0
7.6
1.6
.6
6.4
.7
1.8
.4
State Employment Security Services—159
Employment
Service
The U.S. Employment Service (ES), like the UI, is a joint Federal-State program
that has operated for decades. As with the UI, the Federal Government provides general operating guidelines, oversight, and financing, and the States operate the Service.
But the ES goal, mission, objectives, and operations are quite different from those of
the UI. The basic goal of the Employment Service is to assist individuals, employed
and unemployed, in finding jobs. In addition, it has several, ancillary missions. But the
emphasis has shifted with the passage of time. Indeed, a few of the institutions, some of
the tasks, and many of the procedures discussed here have been modified and in some
cases eliminated. The following is a brief overview of the ES and how it operated until
1987.
Institutional setting
Although the origin of the ES is open to debate, the name dates from World War I
when the U.S. Employment Service was established to recruit defense workers. The ES
languished following the war, but was given new life by the great depression. The
Wagner-Peyser Act of 1933 formally established the joint Federal-State operation. Then,
the primary function of the ES was to refer unemployed workers to work relief programs. The Social Security Act of 1935 established the unemployment insurance program, which called for a work test as a condition for receiving unemployment payments. The ES was assigned responsibility for administering that test. In the 1940s and
1950s, the ES recruited and referred workers to the defense program, and later helped
veterans and defense workers return to civilian employment.9
The focus of the ES shifted in the 1960s to assisting the disadvantaged and registering welfare recipients. During much of the 1970s, the ES screened, tested, and referred
applicants to Federally sponsored job training programs. The work emphasis shifted in
1982 when many programs were transferred to the Job Training Partnership Act program. At this point, the ES returned to its traditional labor exchange operations, although the registration function (work tests) remained.
Although the ES has returned to its basic roots over the past decade, it has undergone
considerable restructuring and downsizing. In 1979 there were about 2,600 ES offices
and 30,000 State employees.10 By 1987 these numbers had dropped to 1,800 offices
and approximately 26,000 employees.11 More recent figures suggest that the number of
offices and employees continue to shrink.12 Additional State employees work on ES
responsibilities specified by the AFDC and food stamp programs, and Federal labor
contracts.
Since 1933, the Employment Service has had three primary functions: Place people
in jobs (labor exchange), enforce laws and regulations (compliance and enforcement),
and provide statistical data (labor market information). More information is provided
on these functions later. The majority of the funding and staffing is devoted to the first
function, labor exchange, the focus of this study. In 1987, labor exchange activities
accounted for 91 percent of ES trust fund expenditures; compliance and enforcement
accounted for about 4.5 percent; and labor market information accounted for about 4.5
percent. Detailed analysis of ES expenditures in the latter part of the 1970s found a
similar division.13
Funding for the traditional ES program comes from a tax on employers (Federal
Unemployment Tax Act). Monies for some of the nontraditional ES programs, such as
the work test, come from general tax revenue, and when the ES supports other government programs, it is often funded under a reimbursable agreement. These programs
include Disabled Veteran’s Outreach, Local Veteran’s Employment Representative, Targeted Job Placement Credit, Job Training Partnership Act, Food Stamps, Dislocated
Workers, Migrant and Seasonal Farm Workers, Disaster Related Funds, and Trade Assistance Act.
Until 1984, the Department of Labor distributed funds to the individual States to
support their projected ES workload. That is, monies were provided to support the
traditional program staff; administrative, supervisory, and technical (AS&T) staff; labor market information staff; special project staff; and nonpersonal services such as
State Employment Security Services—160
space, utilities, computers, and travel. The budget was built from the ground up. Starting in 1982, the Wagner-Peyser amendments changed the way funds are distributed.
The old method funded the number of staff years needed in each State to cover the
projected workload. The new method distributes the funds based on each State’s civilian labor force and its relative share of the Nation’s unemployment. State officials are
permitted to shift the funds among resource categories as needed. These procedures are
for labor exchange services. Activities that are not part of the basic labor exchange
services, such as those covered by the Trade Assistance Act, are supported through
separate funding.
Following is a discussion of the three primary ES functions.
Labor market information is needed to support many ES activities, local, State, and
national. This function includes the collection of data on employment, hours and earnings, local area unemployment, occupational employment, insured employment, and
wage statistics. The national labor information program also includes the maintenance
of the Dictionary of Occupational Titles, collection of data on permanent plant layoffs,
and closings and maintenance of the job bank openings information file.
The compliance and enforcement function is the responsibility of the national office
staff, but State employees perform the services. Several different activities comprise
this function including certification of housing and pay of migrant workers, and documentation and certification of the need for alien workers.14
Labor exchange services are provided by State Employment Service staff and are
the focus of this study. Five services, which account for the bulk of ES expenditures, are
used here. The first service is employer-centered while the last four are client-focused.
•
•
•
•
•
Identifying job openings and providing employer services
Registering and interviewing job seekers
Counseling applicants
Testing applicants
Matching job applications with openings and referring qualified applicants
to employers.
Identifying job openings, accepting job orders, and contacting employers to solicit
orders (job development) are employer-based activities. Employers are reached by
personal visits, telephone calls, mail, and promotional activities. Job listings are entered into computerized job banks daily. Almost 7.0 million job openings were listed in
fiscal 1987, the last year of the index.
The first step for most clients entering the ES process is an interview during which
an application is taken. The interview identifies the client’s job skills, knowledge, and
interests. Approximately 19.2 million applications were taken in fiscal 1987.
Counseling is available to those who need assistance in choosing a field of work,
who wish to change their occupation, or who have difficulty in holding a job. In fiscal
1987, over 600,000 applicants received job counseling.
Tests are sometimes given to the applicants to assess their skill level. Applicants
who do not have a trade or occupation, or who wish to change occupations, may take
general aptitude, specific aptitude, or general interest tests. The ES administered about
800,000 tests in fiscal 1982, the last year for which test data are available.
The last step in the ES process is job matching and referral of applicants to employers. In fiscal 1987, the ES made about 12.5 million referrals, which resulted in 4.5
million hires or job placements.
National data on ES operations, resources, and outputs is quite varied, ranging from
modest to expansive. There have always been some data, and for some years there have
been massive amounts. The National ES Office has collected and published some output statistics since the late 1930s.15 In the 1970s there was a major move to expand the
data collection effort. Two basic data systems were developed: one to collect program
information, the other to collect resource and cost information. Both systems collected
State Employment Security Services—161
data at the local ES office, aggregated them at the State level, and summarized them at
the national level.
Beginning in 1984, the States were no longer required to supply detailed data to the
national ES office, and it was left up to the individual States as to whether they would
continue to collect and use the information themselves. This resulted in a dramatic
cutback in national information available on operations of the Employment Service.
However, some basic program information continued to be collected by the national
office.
The productivity statistics presented here are a function of these data collection efforts. As discussed later, the indexes are restricted to the 1972-87 period.
Neither the Census of Governments nor the Standard Industrial Classification (SIC)
system separately identify the ES program. The Census of Governments includes the
ES program under the general category of Social Insurance Administration while the
SIC assigns the program to Industry #7361, Employment Agencies. The SIC category
includes:
Establishments primarily engaged in providing employment services except theatrical employment agencies and motion picture casting bureaus.
Establishments classified here may assist either employers or those seeking
employment.16
State administrative offices apparently are assigned to Industry 9441, Administration of Social, Manpower, and Income Maintenance programs.17
Outputs
Employment Service output can be measured in a variety of ways, three of which are
looked at here: Placements, referrals, and services. Placements occur each time that an
employer hires an applicant who is referred by the ES. Referrals take place when an
applicant is identified, matched, and referred to a job opening. Services are the activities that are conducted by the ES staff, which result in job referrals and placements.
Chapter 2 provides the criteria used to evaluate these measures.
Before the three measures are reviewed, two issues common to each measure are
discussed. First, are transactions or individuals counted? And second, how are agricultural activities counted?
Transactions versus individuals. Most Employment Service records present outputs or
services in two ways, first as a transaction and second as an individual. A transaction is
recorded each time a service is rendered, for example, each time a job placement is
made or a test is administered. On the other hand, an individual count is recorded the
first time that an individual receives the service within a fiscal year. Repeat service
within the same fiscal year is not counted.
Some researchers and ES administrators feel that the number of individuals is a
better measure of operations than the number of transactions because the primary ES
effort is expended in the initial registration, counseling, and testing. The rationale is
that a registered applicant taking an intermittent job requires little ES time to reprocess.
Others feel that a transaction count is a better measure for two reasons. First, even
though individuals may not require as much processing work in subsequent visits, as
they did in the first, they do require some effort. One study found that costs were more
directly related to transaction operations than to the number of individuals served.18
Second, the data on transactions are likely to be somewhat more accurate than those
collected on individuals since there is no problem in maintaining and matching transaction records. This can be a real problem with individual records.
The measures and data that are used here focus on transactions. However, casual
examination of the data suggest that there is not much difference in the long-term trends
between the two approaches.
Agricultural versus non-agricultural. The second issue that needs to be addressed is
the role of agriculture, and how it is to be handled. Agricultural work played a major
State Employment Security Services—162
role in ES operations at one time. For example, agricultural placements accounted for
62 percent of total ES placements in 1960. By 1970, placements were evenly divided,
but by 1979 agricultural placements made up only 6 percent of the total. This would not
be an issue except that the work processes used by the ES were quite different for
agricultural and non-agricultural operations.
The Senate budget hearings of 1968 highlighted this issue. The hearings noted that
agriculture placements were in the thousands per staff year for three States while nonagricultural placements were in the hundreds. Very different processes were occurring
when agricultural and non-agricultural placements were made.
The primary problem with agricultural placements was the mass placement, which
relied on crew leaders to recruit workers. Each leader would recruit 30-50 workers,
each of whom was counted as a placement. Mass placements represented 79 percent of
all agricultural placement transactions in 1975, but by 1980, these placements had
dropped to less than 2 percent.
Prior to 1982, the ES maintained separate agricultural and non-agricultural placement records. Thus, it is possible, prior to 1982, to calculate and weight outputs separately for agricultural and non-agricultural operations, and this is the procedure we use
here. Beginning in 1982, no distinction is made in the calculations between agricultural
and non-agricultural data.
Placements
Of the three basic types of output measures, placements is the one that seems to be
most often used. The ES has collected statistics on the number of placements since
1938; it tracks the number of placements monthly; and it includes the number of placements as part of its annual budget justification.
A placement occurs each time that an employer hires an applicant who is referred by
ES. For a placement to be recorded, five steps must take place:
1. A job order form must be prepared before the referral is made;
2. Arrangements must be made with the employer for the referral of an individual
or individuals;
3. The employer must not have specifically requested the individual;
4. A reliable source, preferably the employer, must verify that the individual began
work; and
5. The placement must be recorded on ES forms.19
The basic argument for using placements to measure ES output is that placing individuals in jobs is the role of the Service. Counseling, job bank operations, registration
of workers, and testing clients all support the basic service of getting jobs for individuals. Furthermore, placements are measurable, repetitive, and easily understood. They
are physical measures, and data exist to make the measurements.
The principle arguments against the use of placements are: 1) The ES has other work
tasks in addition to placing individuals; 2) a placement is an outcome, not a final output;
3) the placement data are questionable; and 4) the duration of the job needs to be considered. Each is discussed briefly.
Other work tasks. Some of the ES labor market activities, such as job referrals, are
necessary for placements to occur. Others, such as counseling and testing, may or may
not help in placing individuals. The unemployment insurance program, irrespective of
any other action, requires others, such as registration.
Outcome issues. The outcome issue is more troublesome and it is avoided for several
reasons. First, productivity measurement is concerned with production. Second, outcomes are heavily influenced by external factors, considerations over which managers
have little control. As the ES demonstrates, when the economy is booming, and hiring
is the norm, placing individuals in jobs is relatively easy, but in a declining economy
with high unemployment, placing individuals is difficult. Although the state of the
State Employment Security Services—163
economy is the most crucial external factor for ES operations, there are other considerations too, such as the size of the local labor force, skills requested by employers, and
the availability of the needed skills in the labor force.
The research community generally agrees that ES placements are influenced more
by factors external to ES operations, particularly the state of the economy, than by ES
operations. According to one study:
The State’s unemployment rate consistently yields the highest correlation
(negative) with placements. Adding a measure of new hires and a measure
of the percentage of the work force in lower level jobs boosts the total
variance accounted for to 40 to 45 percent of the total variance in placements. Adding certain additional independent variables to the equation
consistently increases the variance accounted for to 60 to 70 percent of the
total.20
Other studies also found external variables to be extremely important in explaining
placement variance.21
Placement data. The third criticism of placements as a measure of ES output concerns
the placement data themselves. Considerable data are collected on placements, but
there are a number of questions concerning their accuracy.22 The issue is not whether
there are errors, but the magnitude of the errors and whether they introduce bias into the
placement statistics.
Job duration. The fourth criticism is the manner in which placements are calculated.
The issue here is the length of the placement, a proxy measure for its quality. A placement can be for a few hours, several days, or for years.
The ES recognizes the importance of the length of the placement, and since the mid1970s it has collected and published placement statistics by job duration—less than 3
days, 3 to 150 days, and more than 150 days. In 1970, 72 percent of the ES placements
were to jobs expected to last over 150 days. This percentage increased to 76 percent in
1973 and dropped to 71 percent in 1976. Starting in 1991, only the number of individuals placed in jobs expected to last over 150 days were reported separately. In 1990, the
General Accounting Office (GAO) identified 400 local offices that placed most people
in temporary jobs and another 400 offices that placed 3 out of 4 persons in permanent
jobs.23
The question for productivity measurement is whether resource requirements vary
significantly by expected job duration. That is, is more ES employee time spent on
placing a person in a long-term position than in a short-term one? No research was
found which addressed this question at the national level. The statistics presented here
do not separate placements by expected job duration.
Irrespective of the questions concerning the use of placements as the measure of ES
output, it is the most widely used measure, and for that reason a placement transaction
index has been produced. The index reflects the sum of the non-agriculture placement
transactions and the weighted agricultural placement transactions for 1972-82, and the
non-agricultural placement transactions for 1982-87. The two indexes are linked in
1982. The average annual increase over the 1972-87 period is 1.0 percent (table 73).
Referrals
In several respects, referrals are a better measure of ES output than placements for
they are less affected by external economic considerations. A referral takes place each
time an applicant is identified, matched, and referred to an opening. A referral has to
take place before a placement can take place.
Arguments in favor of using referrals as the measure of output are that the number of
referrals is a measure of final output, it is measurable, it is repetitive, and it reflects the
work of ES employees. Using referrals as the measure of output avoids the verification
and some of the external economic considerations that are endemic to placements. A
State Employment Security Services—164
count of transaction referrals captures the additional work needed for hard-to-place
individuals and for repeat processing of workers in temporary or intermittent jobs. Not
all referrals, however, lead to jobs, and not all referrals are to job openings. The ES
also refers applicants to training and to support services such as counseling and testing.
Arguments against the use of referrals as the measure of ES output are fourfold.
First, they do not capture all ES services. That is, intake, testing, and counseling are not
explicitly counted. For example, an ES client can be counseled or tested without being
referred to a job. Second, referrals are affected by external economic conditions, like
placements, although not to as great an extent. Third, there are no estimates of the
accuracy of the referral data. It is possible, for example, to manipulate the referral
statistics by referring individuals to multiple job openings. Fourth, national data for
some years are missing and/or are incomplete.
To construct a referral index for 1972-87, a proxy count was used for several years.
Referral data are not available for 1972-75 and 1982-86. For these years the placement
rate of change is used to estimate referral trends. For 1986-87, referral data are again
available and used. By linking the individual segments of the referral output index, that
is, 1972-75, 1975-82, 1982-86, and 1986-87, creates an index covering 1972-87. The
average annual increase over the 1972-87 period is 0.9 percent (table 73).
Services
Exchange services, the third possible ES output measure, is comprised of two basic
groups of services, those for employers and those for applicants. Employer services
include labor exchange and technical assistance. Applicant services include intake,
counseling, testing, and referrals. Law and/or ES regulation require both employer and
applicant services, and all are integral parts of ES operations. The suggested service
measures for employers and applicants are presented in table 71.
Employer services
For purposes of discussion, employer services are divided into two parts, technical
services and labor exchange support services. Technical services help employers by
providing information on legislation affecting their activities, developing occupational
tests for employment use, assisting with recruitment, and helping with affirmative action planning. Employer technical services are a small part of ES operations. In 1975,
the last year for which data are available on ES staff by individual activity, they accounted for only 0.5 percent of all ES employees. Given their apparent small role and
the lack of data to measure their output, they are not considered further.
Labor exchange support services focus on making job development contacts and
taking job orders. Job development is the process whereby ES staff obtains job listings
from employers. In some cases this requires general canvassing of potential employers
for actual and potential job openings; in other instances it requires searching for a specific job for a specific individual.
Job orders is the process of taking and listing of employer initiated requests for job
applicants. Employment Service staff is obligated to process job orders regardless of
whether suitable applicants are available to fill them. According to data for 1975, the
taking of job orders accounted for about 9 percent of all ES employment. However,
during times of economic growth, an even greater proportion of ES staff is used to take
job orders.
A number of different measures have been proposed to track employer services (table
71). The only measure used in this bulletin is the number of job orders listed with ES.
It is measurable, relatively straightforward, data have long been collected and are readily
available, and it is the primary activity of the employer services.
The average annual change in job openings between 1972 and 1987 is -0.4 (table
72). The trends evident in this index are a reflection of the business cycle and the
downsizing of ES operations. The last year of the index, 1987, the ES recorded about
7 million job openings.
State Employment Security Services—165
Table 71. Candidate output measures for ES services
Service
Measures with variations
Employer services
Labor exchange
Number of job orders taken
Number of job development contacts
Technical assistance
Number of employers assisted
Number of services rendered
Number of tests developed
Number of requests handled
Applicant services
Intake
Number of active applicants
Number of applications taken
New application
Renewal application
Partial application
Number of individuals registered
New application
Renewal application
Partial application
Counseling
Number of people counseled
In a group setting
In a one-on-one session
Number of counseling transactions
In a group setting
In a one-on-one session
Testing
Number of people tested in a group session
Proficiency test
Skills test
Number of people tested in a one-on-one
session
Proficiency test
Skills test
Referrals/placements
Number
Number
Number
Number
of referral transactions
of individuals referred
of placement transactions
of individuals placed
Applicant services
Applicant services consist of four activities: Intake, counseling, testing, and referrals.
Intake, the first step in the application process, determines the applicant’s job skills,
education, knowledge, interests, and work experience; checks past applications; and
refers the individual to the next step in the process. As part of the job application
process the person may be referred to counseling or a training program.
The ES is required by law to register and provide service to several categories of
applicants including many income maintenance recipients. Unemployment insurance
applicants, for example, are required to register with the ES but the applicant may be
attached to a former employer and thus not available for work, or the applicant may
have job skills for which the ES has no requests. Nevertheless, registration is still
required.
The obvious output measure for the intake process is the number of intakes, and that
is the basic measure used here. It is measurable, easily understood, reflects the work of
the activity, and data exist with which to calculate the measure.
State Employment Security Services—166
The ES uses several different variations of the basic application. There is the first
time (new application) and reapplication (renewal application). New applications are
further distinguished between complete applications and partial applications. There are
procedural variations for processing each of these types with different labor requirements for each. Estimates show that the partial new application and the renewal application take about half the labor time required to process an entirely new application.
Hence, the need to differentiate the different types of intakes.
One problem in differentiating intake is that the required data are available for only
part of the measured period. The procedure used here is dictated by data availability.
The 1972-75 segment reflects the number of individuals filing new applications. The
1975-82 period differentiates new applications, new partial applications, and renewal
applications, and weights each to calculate an annual index. The 1982-87 segment
reflects the number of active applicants in the ES system. These three segments are
linked to create a single applicant intake index.
There is another potential problem in measuring the intake count in recent years, and
that is the use of group intakes. Group intake is a process whereby ES staff takes
applications from two or more individuals at the same time. A 1989 study found that
more than one-quarter of all offices used group intake to process the majority of their
applicants.24 The reason for using the group method is that ES staff can process about
25 percent more applicants in a given time period. But, there is a question as to whether
the intake procedure is as accurate and complete in the group process. For productivity
calculations, the best approach would be to separate group and individual intake data,
but the data are lacking to make such a calculation so no such distinction is made here
between the two approaches.
In 1987 there were about 19 million active applications in the Nation’s ES offices.
The average annual increase in applicant registration between 1972 and 1987 was 1.1
percent. But there are two different trends in the index. Between 1972 and 1980, it
increase 4.9 percent annually, but over the 1980-87 period, it decreased 3.1 percent
annually. These changes in the overall application index reflect fluctuations in the
business cycle. More importantly, they reflect changes in law that require income maintenance recipients to register and look for work in order to draw benefits (table 72).
Counseling, another step in the application process, is for those who are not yet ready
for a job. Counseling provides employment and occupational information to applicants,
helps them interpret test results, and helps them develop employment strategies.
Counseling is included as a separate output measure to capture the additional work
effort needed to process the “less job ready” applicant.25 This type of applicant was a
focus of ES service during the 1972-82 period, and considerable resources were expended on these applicants in response to policy concerns and directions.
Applicants are counseled individually or as part of a group. When group sessions
are conducted, it is preferable, for productivity measurement, to count the number of
sessions conducted rather than the number of individuals attending the session. This is
because the time needed by ES employees to conduct a session is more closely related
to the number of sessions than to the number of persons counseled.
But, in the case of the ES it is not possible to separate group from individual counseling. Data on group sessions are reported for only about half the measured years and
data are lacking with which to calculate weights which are needed to combine individual and group session counseling. The data presented here are the number of times
a counseling service is rendered. No distinction is made between individual counseling
sessions or group sessions. The resulting index probably slightly overstates the work
by the ES counselors.
Irrespective of which measure is used, counseling dropped dramatically between
1972 and 1987 (table 72). The average annual decrease, using the number of counseling interviews, was 6.4 percent. In 1972, approximately 2.5 million counseling interviews were conducted in ES offices. By 1987 the figure had dropped to less than a
million. Sharp declines were reported in 1974 (21 percent), 1975 (19 percent), and
State Employment Security Services—167
1982 (40 percent). Counseling has been hard hit by the cutback in ES funding. One
study found that the number of full- or part-time State counselors declined by one third
between 1981 and 1987. In 1987, six States had no staff designated as counselors.26
Testing, another applicant service, is often conducted in conjunction with counseling.
The ES administers a number of different tests in screening applicants and helping them
decide on the type of job or job training in which they are interested. The output measure for testing should be the number of tests administered or some variation of this
measure. To compute a testing index, data should be collected on the number of tests
given, labor weights should be developed for each test, and a weighted index computed.
For a number of years the Department of Labor collected statistics on the number
and types of tests given. But no data were collected on the labor required to administer
the tests. Hence, it is not possible to calculate a weighted test index for even part of the
period. Furthermore, no national test data of any type were collected after 1982. The
only usable data are from 1972-82 on the number of tests (transactions) administered
(table 72).
Although there is some question concerning the data, there is no question concerning the drop in the number of tests administered. The average annual decrease from
1972-82 is 6.1 percent. A 1987 study found that most ES offices had dramatically cut
back or eliminated testing.27
Referral of the individual to a job opening is the final step in the applicant process . The
output measure for this step is the number of referrals, the same measure discussed
earlier in this chapter as one of the basic output measures. As already noted, this measure is physical, repetitive, and reflects the work of ES employees. There are data
problems, however, but probably no worse than those of the other services.
The average annual increase in the number of referrals is 0.9 percent between 1972
and 1987, but the rate is heavily influenced by the period examined. As with the other
outputs, it is driven by changes in the economic situation and in government law and
funding (table 72).
Service summary. To compute the service-based output index, five individual service
or activity indexes are weighted and combined: Job orders, intake applications, counseling, testing, and referrals. One measure represents employer services: the number of
job orders. Applicant services are measured by four outputs: Applications, counseling,
testing, and referrals.
As discussed in chapter 2, indexes are combined with base year weights, which usually are revised every 5 years. But in the case of the ES, service labor data are available
for only 1 year, 1980. These data show that job orders consumed 11 percent of the
direct ES labor; applications used 34 percent; counseling, 7 percent; testing, 3 percent;
and referrals, 45 percent. There is one additional complication in computing the service index: data on the number of tests administered are available for 1972-82 only.
Hence, the testing output series is not included in the overall index for 1982-87. This
required that separate indexes and labor weights be calculated for 1972-82 and 198287 and that the two indexes be linked in 1982 (table 72).
The results of the service output index calculations show no change over the entire
1972-87 period. This is in marked contrast to the individual services or activities,
which register very different rates of change over the period table 72.
The total index reflects the individual activities weighted by their importance as
measured by their labor input. Those activities with positive rates of growth, intake
applications and referrals, account for about 80 percent of the labor index. Consequently, these activities shape the index.
Movements within the 1972-87 period differ depending on the activity. Counseling
and testing, for example, show a fairly steady drop reflecting the cut back of these
activities. Others, such as applications and referrals, rise and fall throughout the period
reflecting changes in the economy and revisions to the laws and regulations governing
State Employment Security Services—168
ES operations (table 72).
Table 72. Comparison of individual ES service outputs and total ES service output, 1972-87
(1972 = 100)
Year
Recommended output
series
Job
Weighted
openings Applications Counseling Testing Referrals
total
1972 .................
1973 .................
1974 .................
1975 .................
1976 .................
1977 .................
1978 .................
1979 .................
100.0
121.4
117.0
94.7
110.4
131.2
129.7
128.9
100.0
122.6
127.0
131.8
131.6
135.6
131.9
136.6
100.0
103.8
81.8
66.5
63.9
67.5
70.3
70.1
100.0
103.4
96.3
69.1
66.4
72.8
75.8
80.0
100.0
118.7
126.5
109.9
116.7
136.8
150.8
152.7
100.0
117.6
117.8
106.9
111.1
123.8
128.6
130.8
1980 .................
1981 .................
1982 .................
1983 .................
1984 .................
1985 .................
1986 .................
1987 .................
110.8
102.7
83.7
86.4
89.1
91.8
94.5
94.8
147.0
143.0
120.2
119.7
119.3
118.9
118.5
118.0
73.9
74.5
44.4
43.1
41.8
40.5
39.2
36.9
79.4
79.9
53.3
146.8
140.9
108.6
111.7
114.7
117.7
120.8
115.0
129.6
125.1
97.9
99.2
100.5
101.9
103.2
100.4
Average annual
rate of change:
1972-87 ............
1974-86 ............
1974-79 ............
1979-86 ............
-0.4
-1.8
2.0
-4.3
1.1
-.6
1.5
-2.0
-6.4
-5.9
-3.0
-8.0
0.9
-.4
3.8
-3.3
0.0
-1.1
2.1
-3.3
-3.6
Three output indexes have been computed as part of this examination of the ES:
Placements, referrals, and services. Placements are most often used as the measure of
output. It is the primary goal of the ES. However, there are several problems with using
placements as the measure of output. First, there are other ES activities, and second,
placements are more heavily influenced by economic considerations than the two other
measures. Placements are categorized as an outcome, not a measure of final output.
The number of referrals is another potential measure of output. It is a final output of
the Employment Service and is not affected by external economic considerations to the
degree that placements are. However, there are other ES outputs that are not counted in
a measure of referrals.
The third measure of output is the service-based index. The service-based output
index captures all major services of the ES, many of which are required by law. Job
orders must be taken regardless of whether suitable job applicants are available, and
applications must be taken regardless of the number of job vacancies or the job readiness of the applicant. Furthermore, ES services have changed as the ES mission has
changed. Although the ES has always referred individuals to jobs, the focus in the
1960s and early 1970s was on recruiting and placing the disadvantaged. With this
emphasis went job testing and client screening (other organizations conducted job training). The service-based output index captures the shifts in mission to a greater extent
than do the other measures. It is for this reason that it is used as the measure of ES
output.
Comparison of the indexes for the 1972-87 period shows that the average annual
rate of change was 1.0 for placements, 0.9 for referrals, and 0 for services. For the
peak-to-peak output period of 1974-79, the figures were 4.6, 3.8, and 2.1, respectively.
But for the 1979-86 peak-to-peak periods, each index registered the same average annual decrease of 3.3 percent. It is interesting, but not surprising, that the three output
State Employment Security Services—169
index peaks occurred in the same years for each, and the trends moved in the same
direction. This comparison suggests that the years selected for comparison will have a
greater affect on the rates of change than does the output measure selected (table 73).
Table 73. Comparison of three Employment Service output
indexes, 1972-87
(1972=100)
Year
Labor inputs
Placement
Referral
Service
1972 ...................
1973 ...................
1974 ...................
1975 ...................
1976 ...................
1977 ...................
1978 ...................
1979 ...................
100.0
118.7
126.5
109.9
115.9
138.1
155.2
158.1
100.0
118.7
126.5
109.9
116.7
136.8
150.8
152.7
100.0
117.6
117.8
106.9
111.1
123.8
128.6
130.8
1980 ...................
1981 ...................
1982 ...................
1983 ...................
1984 ...................
1985 ...................
1986 ...................
1987 ...................
142.2
133.1
112.6
115.7
118.9
122.0
125.2
116.4
146.8
140.9
108.6
111.7
114.7
117.7
120.8
115.0
129.6
125.1
97.9
99.2
100.5
101.9
103.2
100.4
Average annual
rate of change:
1972-87 ..............
1974-86 ..............
1974-79 ..............
1979-86 ..............
1.0
-.1
4.6
-3.3
0.9
-.4
3.8
-3.3
0.0
-1.1
2.1
-3.3
Labor dominates Employment Service resource inputs. Although current statistics
are lacking, a 1980 study reported that labor consumed about 85 percent of the ES
budget.28 Sizable sums have been spent on communications and computers in recent
years, but the available data and anecdotal information suggests that labor remains the
primary expenditure of the ES.
The same two labor measures are recommended for calculating State and local government labor productivity as noted in chapter 2: total employment and full-time-equivalent employment. The measure that is used here is the number of State ES positions. A
position is equivalent to an FTE, and positions were used to budget and account for ES
employees until 1982. No national data are available on the current number of ES
employees.
Federal funds supported a fixed number of State employee positions prior to 1983.
In 1982 the Federal Government cut back Employment Service funding, and most ES
funding was shifted to a formula base which reflected each State’s civilian labor force
and its relative share of national unemployment. No longer were Federal funds allocated to support a specific number of State employees. The end result of these actions
was to cut back the number of ES employees, and shortly thereafter the Federal Government abolished State position reporting.
Data to construct the ES employment index for 1972-82 are drawn from the Federal
budget. Starting with 1984, the Federal Government no longer required the States to
report the number of ES positions. Thus, the data for the index for 1982-87 had to be
taken from several different sources including U.S. Congressional appropriation hearings, State reports, and from a 1987 U.S. General Accounting Office survey of the
States. The GAO data are the last collected on ES positions so far as is known, and it is
for this reason that the ES labor and productivity indexes stop with 1987.
State Employment Security Services—170
From 1972-82, the Federal Government funded about 30,000 ES positions. In some
years the number was slightly above this figure, in other years, particularly the mid1970s, it dropped below this number. Overall there was little change until 1982. In
1982, ES funding cuts led to a cut of about 20 percent in the number of ES positions.
The overall 1972-87 average annual change was -0.9 percent for the labor index (table
74).
Productivity trends
Employment Service output per employee year increased 1.0 percent annually between 1972 and 1987, the only years for which data are available. However, there were
two distinct periods of change. Measuring peak-to-peak output periods for this sector,
from 1974-79 the average annual increase in output per employee year was 2.2 percent.
But from 1979-86, a period of considerable turmoil in the ES, there was an average
annual decrease of 1.2 percent. These data reflect the service-based output measure
(table 74).
Earlier in the discussion three different measures of output were presented: Placements, referrals, and service. Each can be used to compute productivity trends, and the
results are somewhat different. For the entire 1972-87 period, the average annual increase of output per employee year for placements was 2.0 percent, for referrals it was
1.9 percent, and for service, as noted above, it was 1.0 percent.
Equally interesting is the change among the different time periods. For 1974-79, the
figures were 4.6 percent, 3.9 percent and 2.2 percent, respectively. But for 1979-86, the
average annual change for each of the three measures was exactly the same, -1.2 percent.
Table 74. Employment Service productivity index, 1972-87
(1972 = 100)
Year
Servicebased
output
Employee
(FTE)
Output
per
employee
1972 .............................................
1973 .............................................
1974 .............................................
1975 .............................................
1976 .............................................
1977 .............................................
1978 .............................................
1979 .............................................
100.0
117.6
117.8
106.9
111.1
123.8
128.6
130.8
100.0
101.5
99.7
92.2
91.3
98.1
99.6
99.3
100.0
115.8
118.2
116.0
121.7
126.3
129.1
131.7
1980 .............................................
1981 .............................................
1982 .............................................
1983 .............................................
1984 .............................................
1985 .............................................
1986 .............................................
1987 .............................................
129.6
125.1
97.9
99.2
100.5
101.9
103.2
100.4
99.3
99.3
79.5
82.1
82.1
83.7
85.4
87.0
130.5
125.9
123.2
120.8
122.4
121.6
120.8
115.4
Average annual
rate of change:
1972-87 ........................................
1974-86 ........................................
1974-79 ........................................
1979-86 ........................................
0
-1.1
2.1
-3.3
-0.9
-1.3
-.1
-2.1
1.0
.2
2.2
-1.2
State Employment Security Services—171
Summary and
Conclusions
This chapter discussed the measurement of State government employment security
services, that is, the Unemployment Insurance program and the Employment Service.
These two services are closely aligned: State government employees operate them
with Federal oversight; they are usually located together; they see many of the same
clients; and State employees are often shifted among the two programs in response to
workload dictates. But in some respects the two services are quite different. The UI is
much larger than the ES when measured by expenditures and employment and it is an
income maintenance program while the ES is a job service program. There are detailed
data on most aspects of the UI, but only limited information on ES operations. Finally,
there have been great changes in the ES, particularly over the past decade.
In fiscal 1992, the States and the Federal Government spent slightly more than $41
billion on the two employment security operations discussed here. Most—98 percent—
was spent by the UI, and most of this was in the form of transfer payments, that is, UI
stipends. Administrative expenditures were about $3.5 billion for the UI and ES; the UI
received about two thirds of this amount. The number of positions or FTE employees
was about 66,000 in 1987, the last year for which detailed employment data are available. In 1987, the UI employed about 60 percent of the staff and the ES about 40
percent.
Unemployment Insurance. As noted earlier, the UI is a well-established and well-documented State-Federal program. Three considerations drive the change in UI output:
Fluctuations in the economy, growth in the Nation’s labor force, and changes in the laws
that specify the benefit payment period and coverage of employees. Between 1964 and
1992, the average annual increase in UI output was 2.5 percent with very large fluctuations in output in some years. Labor input, which follows the change in output, also
registered large year-to-year changes. The average annual increase in labor was 1.3
percent per year. Labor productivity increased 1.2 percent per annum during this period (table 69).
Employment Service. The ES is an even older program, but much smaller and less well
known. The ES, like the UI, is shaped by fluctuations in the economy and by changes in
government law and funding. ES output is effectively constrained by its resources.
Between 1972-87, the only years for which ES data are available, labor input dropped
0.9 percent per year while output was essentially unchanged using the service-based
measure. Labor productivity increased 1.0 percent per year, on average, throughout
this period (table 74).
Employment Security Services. Data for the UI and the ES have been combined to
create a single index for State employment security operations. The two services share
many of the same facilities and see many of the same clients. Also, since their workload
often moves in opposite directions, that is, the UI workload is contracyclical while the
ES tends to be cyclical, employees in the two services are sometimes shifted to work in
the service with the immediate need. Some employees are cross-trained. For productivity analysis, these reasons support the examination of the two services as a single
entity.
Unfortunately, the number of years that can be measured is limited. Although the UI
data cover 28 years, the ES covers only 15 years, and nothing more recent than 1987.
Thus, the index is restricted to 1972-87.
To compute the employment security index, the outputs of the two services are combined using base year labor weights to generate a new output index. The importance of
each output is reflected in the overall output index by the amount of labor used to
produce the output. The base years are 1972, 1977, and 1982 in keeping with the years
used in the other calculations. The labor index is the index of the sum of the employment of the two services. The labor productivity index is calculated by dividing the
output index by the input index.
State Employment Security Services—172
The results of these calculations show an average annual increase in labor productivity of 0.9 percent between 1972 and 1987. Output increased 0.7 percent while labor
dropped 0.2 percent per year (table 75). The most notable features of these calculations
are the year-to-year fluctuations. (See chart 12.)
Table 75. State government employment security output, labor
input and labor productivity indexes, 1972-87
(1972 = 100)
Year
Output
Labor
input
Productivity
1972 ...................
1973 ...................
1974 ...................
1975 ...................
1976 ...................
1977 ...................
1978 ...................
1979 ...................
100.0
102.5
106.5
137.6
151.4
141.0
123.6
119.7
100.0
95.9
92.4
106.4
124.7
124.6
116.4
115.7
100.0
106.9
115.3
129.3
121.5
113.1
106.2
103.4
1980 ...................
1981 ...................
1982 ...................
1983 ...................
1984 ...................
1985 ...................
1986 ...................
1987 ...................
136.8
132.9
137.3
147.5
117.2
117.5
115.7
110.7
115.7
119.6
113.1
119.3
104.9
102.9
96.0
97.1
118.3
111.1
121.4
123.6
111.8
114.2
117.2
113.9
Average annual
rate of change:
1972-87 ..............
0.7
-0.2
0.9
Chart 12. Employment security output, labor, and labor productivity
indexes, 1972-87
Index, 1972=100
160
140
120
100
80
60
40
20
0
1972
Output
1975
Productivity
1978
1981
Labor
1984
160
140
120
100
80
60
40
20
0
1987
State Employment Security Services—173
Endnotes
1
For further details see, U.S. Bureau of Labor Statistics, Measuring Productivity in State and
Local Government, Bulletin 2166, December 1983, pp. 42-51; Donald M. Fisk, “Modest productivity gains in State Unemployment Insurance Service,” Monthly Labor Review, January 1983,
pp. 24-27; and U.S. Bureau of Labor Statistics, Productivity Measures for Selected Industries
and Government Services, Bulletin 2440, Washington: U.S. Government Printing Office, March,
1994, pp. 110-11 and 127.
2
Bruce H. Dunson, S. Charles Maurice and Gerald P. Dwyer, Jr., The Cyclical Effects of the
Unemployment Insurance (UI) Program, Unemployment Insurance Occasional Paper 91-3,
Washington: U.S. Department of Labor, 1991.
3
The Budget for Fiscal Year 1995, Appendix A, Washington: U.S. Government Printing Office,
1995, p. 613.
4
The Budget for Fiscal Year 1994, Appendix A, Washington: U.S. Government Printing Office,
1994, pp. 791-92.
5
U.S. Office of Management and Budget, Standard Industrial Classification Manual, 1987,
Washington: U.S. Government Printing Office, 1987, p. 413.
6
For additional discussion, see the latest Unemployment Insurance Quality Appraisal Results,
an annual publication of the U.S. Department of Labor, Employment and Training Administration.
7
For example, see Paul L. Burgess and Jerry L. Kingston, An Incentives Approach to Improving
the Unemployment Compensation System, Kalamazoo, Michigan: W.E. Upjohn Institute for Employment Research, 1987, pp. 133-58.
8
All staff year counts exclude special program personnel such as those assigned to the trade
adjustment program.
9
Henry P. Guzda, “The U.S. Employment Service at 50: It Too had to Wait Its Turn,” Monthly
Labor Review, June 1983, pp. 12-19.
10
Employment and Training Report of the President, 1980, Washington: U.S. Government Printing
Office, 1980, p. 57.
11
“Employment Service: Variations in Local Office Performance,” GAO/HRD 89-116BR, Washington: U.S. General Accounting Office, August 1989, p. 1.
12
Congressional hearings in 1989 suggest that the ES staff had shrunk to 17,000. But these
numbers may not be comparable to those listed elsewhere in this bulletin because ES staff reporting was terminated in 1984 as discussed later. However, there is no doubt that the ES staff had
shrunk dramatically in the 1980s. See United States Congress House Committee on Ways and
Means, Subcommittee on Human Resources, Reforming the Unemployment Compensation System, May 24, 1989, Washington: U.S. Government Printing Office, 1990, pp. 101, 151, and 218.
13
Charles K. Fairchild, A Performance and Needs Based Methodology for Allocating Employment Service Grants, Cambridge, Massachusetts: Abt Associates, 1980, pp. 12-13.
14
“The Impact of Employment Service Regulations on the ES Network, Phase I Findings,” Booz
Allen & Hamilton, Inc., 1980, pp. V17-V24.
15
Historical Statistics of Employment Security Activities, 1938-66, Washington: U.S. Bureau of
Employment Security, 1968.
16
U.S. Office of Management and Budget, Standard Industrial Classification Manual 1987,
Washington: U. S. Government Printing Office, 1987, p. 364.
17
U. S. Office of Management and Budget, Standard Industrial Classification Manual 1987,
Washington: U. S. Government Printing Office, 1987, pp. 412-3.
18
Edward E. Cavin and Frank P. Stafford, The Journal of Human Resources, “Efficient Provision of Employment Service Outputs: A Production Frontier Analysis,” 1985, p. 495.
State Employment Security Services—174
19
“Glossary of Program Terms and Definitions,” Washington: U.S. Employment and Training
Administration, 1978.
20
John P. Campbell, “Comments” in David W. Stevens and others, Specification and Measurement of Productivity in USES, Washington: U.S. Employment Service, December, 1980, p. 188.
21
Charles K. Fairchild, A Performance and Needs Based Methodology for Allocating Employment Service Grants, Cambridge, Massachusetts: Abt Associates, 1980, and Charles O. Thorpe,
Jr. and Richard S. Toikka, Determinants of State Employment Service Productivity, Washington:
The Urban Institute, March, 1979.
22
For a discussion of the different types of errors see Measuring Productivity in State and Local
Government (BLS Bulletin 2166), Washington: U.S. Government Printing Office, December
1983, pp. 78-80.
23
“Employment Service: Leadership Needed to Improve Performance,” Washington: U.S. General Accounting Office, October 16, 1990 testimony, GAO/T-HRD-9-4, p. 1.
24
Employment Service: Variations in Local Office Performance, GAO/HRD, 89-116BR, Washington: U.S. General Accounting Office, August 1989, p. 25.
25
Neil S. Weiner, John H. Powel, and C. Michael Rahm, The United States Employment Service:
A Conceptual Model of Outputs, Values and Illustrative Estimations, Vol. II, Arlington, Va:
Boeing Computer Services, 1976, p. B-11.
26
“Employment Service: Variations in Local Office Performance,” GAO/HRD, 89-116BR, Washington: U.S. General Accounting Office, August 1989, p. 29.
27
“Employment Service: Variations in Local Office Performance,” GAO/HRD, 89-116BR, Washington: U.S. General Accounting Office, August, 1989, p. 29.
28
Charles K. Fairchild, A Performance and Needs Based Methodology for Allocating Employment Service Grants: Cambridge, Massachusetts: Abt Associates 1980, p. 39.
State Employment Security Services—175
Chapter 6. Bibliography
General
Balk, Walter L., Geert Bouckaert and Kevin M. Bronner, “Notes on the Theory and
Practice of Government Productivity Improvement,” Public Productivity and Management Review, Vol. XII, No., 2, Winter 1989.
Burkhead, Jesse and Patrick J. Hennigan, “Productivity Analysis: A Search for Definition and Order,” Public Administration Review, January/February, 1978, pp. 34-40.
Bouckaert, Geert, “Public Productivity in Retrospective,” in Marc Holzer (editor), Public
Productivity Handbook, New York: Marcel Dekker, Inc., 1992, pp. 20-29.
Economic Report of the President, Washington: U.S. Government Printing Office, 1995.
Forte, Darlene, “Measuring Federal Government Productivity,” in William F. Christopher and Carl G. Thor, Handbook for Productivity Measurement and Improvement,
Productivity Press, 1993, pp. 7-3.1.
Goudriaan, R., Hans de Groot, and Frank van Tulder, “Public Sector Productivity: Recent Empirical Findings and Policy Applications,” Proceedings of the 41st Congress of
the International Institute of Public Finance, pp. 193-209, Detroit: Wayne State University Press, 1987.
Halachmi, Arie, and Marc Holzer, “Introduction: Toward Strategic Perspectives on Public
Productivity,” in Marc Holzer and Arie Halachmi (ed), Strategic Issues in Public Sector Productivity, San Francisco: Jossey Bass, 1986, pp. 5-14.
Kunze, Kent, Mary Jablonski and Virginia Klarquist, “BLS Modernizes Industry Labor
Productivity Program,” Monthly Labor Review, July 1995, pp. 3-12.
Levitt, Malcolm and Michael Joyce, “Measuring Output and Productivity in Government,” Discussion Paper No. 108, London: National Institute of Economic and Social
Research, 1986.
Mark, Jerome A., “Progress in Measuring Productivity in Government,” Monthly Labor Review, December 1972, pp. 3-6.
Murray, Richard, “Measuring Public-Sector Output: The Swedish Report,” pp. 518-19,
in Ziv Griliches (editor), Output Measurement in the Service Sectors, Chicago: The
University of Chicago Press, 1992.
National Research Council, Measurement and Interpretation of Productivity, Washington: National Academy of Sciences, 1979.
U.S. Bureau of Labor Statistics, BLS Handbook of Methods, Bulletin 2414, Washington: U.S. Government Printing Office, September 1992, pp. 78-98.
Bibliography—176
U.S. Bureau of Labor Statistics, Employment and Earnings, monthly publication.
U.S. Bureau of Labor Statistics, Measuring Productivity in State and Local Government, Bulletin 2166, Washington: U.S. Government Printing Office, December 1983.
U.S. Bureau of Labor Statistics, Productivity Measures for Selected Industries and
Government Services, Bulletin 2440, Washington: U.S. Government Printing Office,
March 1994.
U.S. Congress, Joint Economic Committee, Productivity in the Federal Government,
Washington: U.S. Government Printing Office, 1979.
U.S. General Accounting Office, The Federal Role in Improving Productivity—Is the
National Center for Productivity and Quality of Working Life the Proper Mechanism?
May 1978.
Methodology
Bradford, D.F., R.A. Malt, and W.E. Oates, “The Rising Cost of Local Public Services:
Some Evidence and Reflections,” National Tax Journal, Vol. XXII, No. 2 (June 1969),
pp. 185-202.
Bouckaert, Geert, “Measurement and Meaningful Management,” Public Productivity
and Management Review, Vol. XVII, No. 1, Fall 1993, pp. 31-43.
Burkhead, Jesse and Patrick J. Hennigan, “Productivity Analysis: A Search for Definition and Order,” Public Administration Review, January/February 1978, pp. 34-40.
Crandall, Robert W., and Jonathan Galst, “Productivity Growth in the Telephone Industry Since 1984,” Patrick T. Harker(ed), The Service Productivity and Quality Challenge, Boston: Kluwer Academic Publishers, 1995, pp. 391-405.
Diewert, W.E., “The Measurement of Productivity,” Bulletin of Economic Research,
1992, pp. 163-198.
Fielding, Gordon J. and others, Development of Performance Indicators for Transit,
Irvine, California: University of California, 1977.
Fisher, Franklin M. and Karl Shell, The Economic Theory of Price Indices, New York:
Academic Press, 1972.
Fox, William, Size Economies in Local Government Services: A Review, Washington:
U.S. Department of Agriculture, Economic, Statistics and Cooperative Service, 1980.
Greytak, David, Donald Phares, and Elaine Morely, Municipal Output and Performance
in New York City, Lexington, Massachusetts: Lexington Books, 1976.
Grosskopf, S., “Efficiency and Productivity,” chapter 4 in Harold O. Fried, C.A. Knox
Lovell and Shelton S. Schmidt, The Measurement of Productive Efficiency, New York:
Oxford University Press, 1993, pp. 160-193.
Gullickson, William, “Measurement of Productivity Growth in U.S. Manufacturing,”
Monthly Labor Review, July 1995, pp. 13-28.
Holzer, Marc (editor), Public Productivity Handbook, New York: Marcel Dekker, Inc.,
1992.
Bibliography—177
Hjerppe, Reino T., “The Measurement of Real Output of Public Sector Services,” The
Review of Income and Wealth, June 1980, pp. 237-50.
Kunze, Kent, Mary Jablonski, and Virginia Klarquist, “BLS Modernizes Industry Labor Productivity Program,” Monthly Labor Review, July 1995, pp. 3-12.
Murray, Richard, “Measuring Public-Sector Output: The Swedish Report,” in Zvi
Griliches(editor), Output Measurement in the Service Sectors, Chicago: The University of Chicago Press, 1992.
National Research Council, Measurement and Interpretation of Productivity. Washington: National Academy of Sciences, 1979.
Ross, John P. and Jesse Burkhead, Productivity in the Local Government Sector, Lexington, Mass.: Lexington Books, 1974.
Schmidt, Peter and Ann D. Witte, An Economic Analysis of Crime and Justice: Theory,
Methods and Applications, New York: Academic Press, Inc., 1984.
Sherwood, Mark, “Difficulties in the Measurement of Service Outputs,” Monthly Labor Review, March 1994, pp. 11-19.
Shulman, Martha A., “Alternative Approaches for Delivering Public Services,” Urban
Data Service Reports, Vol. 14, No. 10, Washington: International City Management
Association, October 1982.
Tomazinis, Anthony R., Productivity, Efficiency, and Quality in Urban Transportation
Systems. Lexington, Mass.: D.C. Heath and Company, 1975.
Triplett, Jack E., “Robert Gordon’s Approach to Price Measurement,” BLS Working
Paper 101, Washington: U.S. Bureau of Labor Statistics, April 1980.
U.S. Bureau of Labor Statistics, BLS Handbook of Methods, Bulletin 2414, Washington: U.S. Government Printing Office, September 1992.
U.S. Bureau of Labor Statistics, Labor Composition and U.S. Productivity Growth,
1948-90, Bulletin 2426, Washington: U.S. Government Printing Office, December 1993.
U.S. Bureau of the Budget, Measuring Productivity of Federal Government Organizations, Washington: U.S. Government Printing Office, 1964.
U.S. Bureau of the Census, Public Employment, Washington: U.S. Government Printing Office, annual publication.
Usilaner, Brain and Edwin Soniat, “Productivity Measurement,” in George Washnis,
editor, Productivity Improvement Handbook, New York: John Wiley & Sons, 1981.
Whitaker, Gordon P. and others, Basic Issues in Police Performance, Washington: U.S.
National Institute of Justice, 1982.
Wilson, James Q., Bureaucracy: What Government Agencies Do and Why They Do It,
New York: Basic Books, Inc., 1989.
Bibliography—178
Enterprise Services
Electric Power
Government Transactions, “Methodology Papers: U.S. National Income and Product
Accounts,” (BEA-MP-5), Washington: U.S. Bureau of Economic Analysis, November
1988.
Balk, Walter L., editor, Public Utility Productivity, Albany: New York State Department of Public Services, 1975.
Barzel, Yoram, “Productivity in the Electric Power Industry—1929-1955,” Review of
Economics and Statistics, November 1963, pp. 395-408.
Electric Sales and Revenue—1992, U.S. Department of Energy, Energy Information
Administration, 1994.
Financial Statistics of Major U.S. Publicly Owned Electric Utilities 1992, U.S. Department of Energy, Energy Information Administration, 1994.
Gollop, Frank M., and Mark Roberts, “Environmental Regulations and Productivity
Growth: The Case of Fossil-Fueled Electric Power Generation,” Journal of Political
Economy, Vol. 91, August 1983, pp. 654-74.
Gould, Jacob Martin, Output and Productivity in Electric and Gas Utilities—18991942. New York: National Bureau of Economic Research, 1946.
Iulo, William, Electric Utilities—Costs and Performance, Pullman, Washington: State
University Press, 1961.
Kunze, Kent, Mary Jablonski and Virginia Klarquist, “BLS Modernizes Industry Labor
Productivity Program,” Monthly Labor Review, July 1995, pp. 3-12.
Productivity Measures for Selected Industries and Government Services, Bulletin 2440,
Washington: U.S. Government Printing Office, March 1994.
Public Power, annual January/February issue.
Ross, Everett, “Managing Electric Utilities,” in George J. Washnis (editor), Productivity Improvement Handbook for State & Local Government, New York: John Wiley &
Sons, 1980.
Statistics of Publicly Owned Electric Utilities in the United States, selected issues, U.S.
Department of Energy.
U.S. Bureau of Labor Statistics, Measuring Productivity in State and Local Government, Bulletin 2166, Washington: U.S. Government Printing Office, December 1983.
U.S. Bureau of the Census. Public Employment: 1992, Series GE/92-1, U.S. Government Printing Office, Washington, DC, 1994.
U.S. Bureau of the Census. 1987 Census of Governments—Compendium of Government Finances, Bureau of the Census, 1989.
U.S. Bureau of the Census. 1987 Census of Governments—Compendium of Public
Employment, Bureau of the Census, 1989.
U.S. Bureau of the Census. 1992 Census of Governments—Government Finances, 19911992, Bureau of the Census, 1996.
Bibliography—179
U.S. Office of Management and Budget, Standard Industrial Classification Manual,
Washington: U.S. Government Printing Office, 1987.
Wilson, J.W. and Associates, The Measurement of Electric Utility Productivity, Vols. I
and II. Washington: U.S. National Bureau of Standards, 1980.
Natural Gas
Gas Facts-1991 Data, Arlington, Virginia: American Gas Association, 1992.
“Gas and Electric Utilities - SIC 491, 492, 493,” unpublished technical note, Office of
Productivity and Technology, U.S. Bureau of Labor Statistics, July 1989.
Kunze, Kent, Mary Jablonski and Virginia Klarquist, “BLS Modernizes Industry Labor
Productivity Program,” Monthly Labor Review, July 1995, pp. 3-12.
Natural Gas 1992: Issues and Trends, Washington: U.S. Energy Information Agency,
March 1993.
Stanfill, Ira C., “Gas Utilities,” in George J. Washnis (editor), Productivity Improvement Handbook for State & Local Government, New York: John Wiley & Sons, 1980.
U.S. Bureau of Labor Statistics, Productivity Measures for Selected Industries and
Government Services, Bulletin 2440, Washington: U.S. Government Printing Office,
March 1994.
U.S. Bureau of the Census, Government Finances: 1991-1992, U.S. Government Printing
Office, Washington, DC, 1996.
U.S. Bureau of the Census, Public Employment: 1992, Series GE/92-1, U.S. Government Printing Office, Washington, DC, 1994.
U.S. Bureau of the Census, 1987 Census of Governments—Compendium of Government Finances, U.S. Government Printing Office, Washington, DC, 1990.
U.S. Bureau of the Census, 1987 Census of Governments—Compendium of Public
Employment, U.S. Government Printing Office, Washington, DC, 1989.
U.S. Office of Management and Budget, Standard Industrial Classification Manual,
Washington: U.S. Government Printing Office, 1987.
Water Supply
“Additional Federal Aid for Urban Water Distribution Systems Should Wait Until Needs
are Clearly Established,” Washington: U.S. General Accounting Office, November 24,
1980.
Clark, Robert M., “The Safe Drinking Water Act: Its Implications for Planning,” in
David Holz and Scott Sebastian (editors), Municipal Water Systems: A Challenge for
Urban Resource Management, Bloomington: Indiana University Press, 1978.
Clark, Robert M., “Small Water Systems: Role of Technology,” Journal of the Environmental Engineering Division, American Society of Civil Engineers, Vol. 106, No.
EE1, Proceedings Paper 15181, February 1980.
Clark, Robert M., and James I. Gillean, “The Cost of Water Supply Utility Management,” Environmental Modeling and Simulation, Proceedings of the Conference, Cincinnati, Ohio: U.S. Environmental Protection Agency, 1976, pp. 808-813.
Bibliography—180
Clark, Robert M., John A. Machisko and Richard G. Stevie, “Cost of Water Supply:
Selected Case Studies,” Journal of The Environment Engineering Division, February
1979.
“Comparisons of Cost, Manpower Utilization, and Flow in Operation and Maintenance
of Investor Owned Water Companies and Municipal Waste Water Systems,” Washington: U.S. Environmental Protection Agency, June 5, 1979.
“Drinking Water: Compliance Problems Undermine EPA Programs as New Challenges
Emerge,” GAO/RCED 90-127, Washington: U.S. General Accounting Office, June 1990.
“Drinking Water: Stronger Efforts Essential for Small Communities to Comply with
Standards,” Washington: U.S. General Accounting Office, March 1994.
“Drinking Water Program: States Face Increased Difficulties in Meeting Basic Requirements,” RCED93-144, Washington: U.S. General Accounting Office, June 1993.
“Environmental Investments: The Cost of a Clean Environment,” EPA-230-11-90-083,
Washington: U.S. Environmental Protection Agency, November 1990.
Feigenbaum, Susan, and Ronald Teeples, “Public Versus Private Water Delivery: A
Hedonic Cost Approach,” Review of Economics and Statistics, 65, November 1983,
pp. 672-78.
Fox, William, Size Economies in Local Government Services: A Review, Washington:
U.S. Department of Agriculture, 1980.
Gillean, James I. and others, “The Cost of Water Supply and Water Utility Management,” Contract Report 60-03-2071, Cincinnati: U. S. Environmental Protection Agency,
1977.
Hatry, Harry P., et. al., Efficiency Measurement for Local Government Services, Washington: Urban Institute, 1979.
Hatry, Harry P., et. al., Service Efforts and Accomplishments Reporting: Its Time Has
Come. Norwalk, Connecticut: Government Accounting Standards Board, 1990.
Kim, H. Youn and Robert M. Clark, “Input Substitution and Demand in the Water Supply Production Process,” Water Resources Research, Vol. 23. No. 2, pp. 239-44, February 1987.
Rutledge, Gary L., and Christine R. Vogan, “Pollution Abatement and Control Expenditures, 1972-92,” Survey of Current Business, May 1994, pp. 36-49.
Schefter, John E., “Domestic Water Use in the United States, 1960-85,” National Water
Summary 1987 - Hydrologic Events and Water Supply and Use, Washington: U.S. Government Printing Office, 1990.
Solley, Wayne B., Robert R. Pierce and Howard A. Perlman, Estimated Use of Water in
the United States in 1990, USGS Circular 1081, Washington: U.S. Government Printing Office, 1993.
U.S. Bureau of the Census, Census of Governments (1987)—Compendium of Government Finances, Washington: U.S. Government Printing Office, 1990.
Bibliography—181
U.S. Bureau of the Census, Census of Governments (1987)—Compendium of Public
Employment, Washington: U.S. Government Printing Office, 1988.
U.S. Bureau of the Census, City Employment: 1992, Series GE/92-2. Washington: U.S.
Government Printing Office, 1994.
U.S. Bureau of the Census, 1992 Census of Governments—Government Organization,
Washington: U.S. Government Printing Office, 1994.
U.S. Bureau of the Census, Government Finances: 1991-92, Series GF92/5, Washington: U.S. Government Printing Office, 1996.
U.S. Bureau of the Census, Public Employment: 1992, Series GE/92-1, Washington:
U.S. Government Printing Office, 1994.
U.S. Office of Management and Budget, Standard Industrial Classification Manual,
Washington: U.S. Government Printing Office, 1987.
Mass Transit
Bennewitz, Eckhard, “Mass Transit,” in George J. Washnis (editor), Productivity Improvement Handbook for State & Local Government, New York: John Wiley & Sons,
Inc., 1980.
Deakin, B. M., and T. Seward, Productivity in Transportation, Cambridge, England:
Cambridge University Press, 1969.
Fielding, Gordon J., Roy E. Glauthier and Charles A. Love, Development of Performance Indicators for Transit, Irvine, California: University of California, Institute of
Transportation Studies, 1977.
Kendrick, John W., “Productivity Trends in U.S. Transportation Industries,” as cited in
Raymond C. Scheppach, Jr. and L. Carl Woehlcke, Transportation Productivity, Lexington, Mass.: Lexington Books, 1975.
National Transit Summaries and Trends, (other titles include National Urban Mass
Transportation Statistics and Section 15 Annual Report) Washington: U.S. Federal
Transit Administration, selected years.
Scheppach, Jr., Raymond C., and L. Carl Woehlcke, Transportation Productivity, Lexington, Mass.: Lexington Books, 1975.
Transit Fact Book, Washington: American Public Transit Association, selected years.
Transit Operating Report, Washington: American Public Transit Association, selected
years.
U.S. Bureau of Labor Statistics, Measuring Productivity in State and Local Government, Bulletin 2166, U.S. Government Printing Office, December 1983.
U.S. Bureau of Labor Statistics, Productivity Measures for Selected Industries and
Government Services, Bulletin 2440, Washington: U.S. Government Printing Office,
March 1994.
U.S. Bureau of the Census, Public Employment: 1992, Series GE/92-1, Washington:
U.S. Government Printing Office, 1994.
Bibliography—182
U.S. Bureau of the Census, Government Finances: 1991-92, Series GF/92-5, Washington: U.S. Government Printing Office, 1996.
U.S. Office of Management and Budget, Standard Industrial Classification Manual,
Washington: U.S. Government Printing Office, 1987.
Alcoholic Beverages
Wallace, Wanda A., “Mass Transit,” chapter 7, in Harry P. Hatry, et. al., Service Efforts
and Accomplishments Reporting: Its Time Has Come, Norwalk, Connecticut: Governmental Accounting Standards Board, 1990.
Annual Statistical Review, annual issues. Washington: Distilled Spirits Council of the
United States.
Basch, Donald L. and John A. Menge, “Government Enterprise and the Sale of Liquor:
Lessons from New Hampshire’s Experience,” State Government, Vol. 55, No. 2, 1982,
pp. 34-39.
Kunze, Kent, Mary Jablonski and Virginia Klarquist, “BLS Modernizes Industry Labor
Productivity Program,” Monthly Labor Review, July 1995, pp. 3-12.
Public Revenues from Alcohol Beverages, annual issues. Washington: Distilled Spirits
Council of the United States.
“Summary of State Laws & Regulations Relating to Distilled Spirits.” Twenty-eighth
Edition, Washington: Distilled Spirits Council of the United States, December 1993.
U.S. Bureau of Labor Statistics, BLS Handbook of Methods, Bulletin 2414, Washington: U.S. Government Printing Office, September, 1992.
U.S. Bureau of Labor Statistics, Measuring Productivity in State and Local Government, Bulletin 2166, Washington: U.S. Government Printing Office, December 1983.
U.S. Bureau of the Census. Public Employment: 1992, Series GE/92-1, Washington:
U.S. Government Printing Office, 1994.
U.S. Bureau of the Census. Government Finances: 1990-91, Washington: U.S. Government Printing Office, 1993.
U.S. Bureau of the Census. Government Finances: 1991-92, Washington: U.S. Government Printing Office, 1994.
U.S. Office of Management and Budget, Standard Industrial Classification Manual,
Washington: U.S. Government Printing Office, 1987.
York, James D., “Retail Liquor Stores Experience Flat Trend in Productivity,” Monthly
Labor Review, February 1987, pp. 25-29.
Corrections
Tracy L. Snell, Correctional Populations in the United States, 1992, NCJ-146413,
Washington: U.S. Bureau of Justice Statistics, January 1995.
U.S. Bureau of the Census, Census of Governments (1987), Vol. 4, No. 5: Compendium
of Government Finances, Washington: U.S. Government Printing Office, 1990.
U.S. Bureau of the Census, Census of Governments, 1967, Vol. 3, No. 2: Compendium
of Public Employment, Washington: U.S. Government Printing Office, 1969.
Bibliography—183
State Adult Correctional
Institutions
U.S. Bureau of the Census, Public Employment: 1992, Series GE/92-1, Washington:
U.S. Government Printing Office, 1994.
Bartollas, Clemens, Correctional Treatment: Theory and Practice, Englewood Cliffs,
N.J.: Prentice-Hall, Inc., 1985.
Blair, Louis H., et. al., Monitoring the Impacts of Prison and Parole Services: An Initial Examination, Washington: Urban Institute, July 1977.
Block, Michael and Thomas Ulen, “Cost Functions for Correctional Institutions,” Charles
M. Gray (editor), The Costs of Crime, Beverly Hills, CA.: SAGE Publications, 1979.
Brown, Edmund J., et. al., 1981 Sourcebook of Criminal Justice Statistics, Washington:
U.S. Government Printing Office, 1976.
Carter, Robert M., et. al., Correctional Institutions, New York: Harper and Row Publishers, 1985.
DiIulio, Jr., John J., Governing Prisons, New York: The Free Press, 1987.
Hackett, Judith C. and others, “Contracting for the Operation of Prisons and Jails,”
Washington: U.S. National Institute of Justice, June, 1987.
Holzer, Marc, Constance Zalk and Jerome Pasichow, “Corrections,” in George J.
Washnis, Productivity Improvement Handbook for State and Local Government, New
York: John Wiley & Sons, 1980.
Innes, Christopher A., “Population Density in State Prisons,” U.S. Bureau of Justice
Statistics Special Report, NCJ-103204, December 1986.
Irwin, John and James Austin, It’s About Time, San Francisco: National Council on
Crime and Delinquency, 1987.
Langan, Patrick A., et. al., Historical Statistics on Prisoners in State & Federal Institutions, Year end 1925-86, Washington: U.S. Bureau of Justice Statistics, May 1988.
McDonald, Douglas, The Price of Punishment: Public Spending for Corrections in
New York, Boulder, Colorado: Westview Press, 1980.
National Planning Association, The National Manpower Survey of the Criminal Justice
System, Vol. 3, Washington: U.S. National Institute of Law Enforcement and Criminal
Justice, September 1978.
1984 Census of State Adult Correctional Facilities, NCJ-105585, Washington: U.S.
Bureau of Justice Statistics, August 1987.
Prisoners—1978, Washington: U.S. Bureau of Justice Statistics, 1979.
Schmidt, P., and Ann D. Witte, An Economic Analysis of Crime and Justice, New York:
Academic Press, 1984.
“Security Staffing in Adult Facilities of the Virginia Department of Corrections,” Richmond: Virginia Department of Planning and Budget, December 1985.
“State and Federal Prisons: Factors That Affect Construction and Operations Costs,”
Washington: U.S. General Accounting Office, May 1992.
Bibliography—184
Tabasz, Thomas F., Toward an Economics of Prisons, Lexington, Mass.: Lexington
Books, 1975.
U.S. Bureau of Justice Statistics, Census of State and Federal Correctional Facilities,
1990, NCJ-137003, Washington: U.S. Government Printing Office, May 1992.
U.S. Bureau of Justice Statistics, Prisoners in State and Federal Institutions on December 31, 1983, NCJ 99861, June 1986.
U.S. Bureau of Justice Statistics, Report to the Nation on Crime and Justice, NCJ105506, Washington: U.S. Government Printing Office, March 1988.
U.S. Bureau of Justice Statistics, Sourcebook of Criminal Justice Statistics, Washington: U.S. Government Printing Office, 1976.
U.S. Bureau of the Census, Census of Governments (1987), Compendium of Public
Employment, Washington: U.S. Government Printing Office, 1988.
U.S. Bureau of the Census, Government Finances: 1989-90, Series GF/90-5, Washington: U.S. Government Printing Office, 1991.
U.S. Bureau of the Census, Public Employment: 1990, Series GE-90-1, Washington:
U.S. Government Printing Office, 1991.
U.S. Office of Management and Budget, Standard Industrial Classification Manual.
Washington: U.S. Government Printing Office, 1987.
U.S. President’s Commission on Law Enforcement and Administration of Justice, Task
Force Report: Corrections, Washington: U.S. Government Printing Office, 1967.
Witte, Ann D., Richard Hofler and others, “Empirical Investigations of Correctional
Cost Functions,” LEAA Grant #78-NI-AX-0059, Unpublished report, 1982.
Jails
Block, Michael K., Cost, Scale Economies and Other Economic Concepts, Washington: American Bar Association, February 1976.
BJS Data Report, 1988, NCJ 116262, Washington: U.S. Government Printing Office,
April 1989.
Census of Local Jails—1983, Washington: U.S. Bureau of Justice Statistics, 1988.
Census of Local Jails—1988: Vol. I. Selected Findings, Methodology and Summary
Tables, Washington: U.S. Government Printing Office, 1991.
Goldfarb, Ronald, Jails: The Ultimate Ghetto, Garden City, N.Y.: Anchor Press, 1975.
Holzer, Marc, Constance Zalk and Jerome Pasichow, “Corrections,” in George J.
Washnis, Productivity Improvement Handbook for State and Local Government. New
York: John Wiley & Sons, 1980.
Innes, Christopher A., Population Density in Local Jails, 1988, BJS Special Report,
NCJ-122299, March 1990.
Jail Inmates, 1987, NCJ 114319, Washington: U.S. Bureau of Justice Statistics, December 1988.
Bibliography—185
Jamieson, Katherine M., and Timothy J. Flanagan, editors, Sourcebook of Criminal
Justice Statistics—1988, U. S. Department of Justice, Bureau of Justice Statistics,
Washington: U.S. Government Printing Office, 1989, p. 605.
Kerle, Kenneth E., and Francis R Ford, The State of Our Nation’s Jails—1982, Washington: National Sheriff’s Association, 1982.
Mancini, Norma, Our Crowded Jails: A National Plight, Washington: U.S. Bureau of
Justice Statistics, January 1988.
Perkins, Craig A., James J. Stephan and Allen J. Beck, Jails and Jail Inmates 1993-94,
NCJ-151651, Washington: U.S. Bureau of Justice Statistics, April 1995.
Reaves, Brian A., “Sheriffs’ Departments 1990,” Washington: U.S. Bureau of Justice
Statistics Bulletin, February, 1992.
Singer, Neil M., and Virginia B. Wright, Cost Analysis of Correctional Standards: Institutional-Based Programs and Parole, Washington: U.S. Law Enforcement Administration Agency, January 1976.
U.S. Bureau of Justice Statistics, Census of State and Federal Correctional Facilities,
1990, Washington: U.S. Government Printing Office, May 1992.
U.S. Office of Management and Budget, Standard Industrial Classification Manual.
Washington: U.S. Government Printing Office, 1987.
Wayson, B. L., et. al., The Cost of Jail Standards Compliance in Washington State,
Washington: American Bar Association, 1975.
Adult Probation and Parole
“Arizona Criminal Justice System Research Projects, Sub-project II, Arizona—Study
of the Cost of Incarceration and Rehabilitation,” Study performed by Lawrence-Leiter
and Company, Kansas City, Missouri (year unknown).
BJS Data Report, 1989, Washington: U.S. Bureau of Justice Statistics, NCJ-121514,
December 1990.
Bemus, Brian, Gary Arling, and Peter Quigley, “Workload Measures for Probation and
Parole,” NIC Grant CU-4, May 1983.
Byrne, James M., “Probation,” Crime File Study Guide, NCJ-104558, U.S. National
Institutes of Justice, 1988.
Cahalan, Margaret Werner, Historical Corrections Statistics in the United States, 18501984, Washington: U.S. Bureau of Justice Statistics, December 1986.
Cohn, Alvin W., and Michael M. Ferriter, “The Presentence Investigation Report: An
Old Saw With New Teeth,” Federal Probation, September 1990.
Correctional Populations in the United States 1985, NCJ-103957, Washington: U.S.
Bureau of Justice Statistics, December 1987.
Flanagan, Timothy J., and Kathleen Maguire (eds), Sourcebook of Criminal Justice
Statistics, 1989, Washington: U.S. Government Printing Office, 1990.
Guynes, Randall, “Difficult clients, large caseloads plague probation, parole agencies,”
Washington: U.S. National Institute of Justice, August 1988.
Bibliography—186
Hartke, Kenneth L., “Work Units in Theory and Practice,” Corrections Today, pp. 6668, February 1984.
Justice Expenditure and Employment in the United States, Washington: U.S. Government Printing Office, selected issues.
Maguire, Kathleen and Ann L. Pastore (editors), Sourcebook of Criminal Justice Statistics—1993, Washington: U.S. Government Printing Office, 1994.
McDonald, Douglas, The Price of Punishment: Public Spending for Corrections in
New York, Boulder, Colorado: Westview Press, 1980,
Michigan Department of Corrections, Annual Report—1987, no date.
National Correction Reporting Program, 1985, NCJ 123522, Washington: U.S. Bureau of Justice Statistics, December 1980.
National Correction Reporting Program, 1990, NCJ 141879, Washington: U.S. Bureau of Justice Statistics, May 1993.
Petersilia, Joan, “Probation and Felony Offenders,” Research in Brief, National Institutes of Justice, March 1985.
Petersilia, Joan and Susan Turner, “Comparing Intensive and Regular Supervision for
High Risk Probationers,” Crime and Delinquency, January 1990, pp. 87-111.
Probation and Parole, 1989, Washington: U.S. Bureau of Justice Statistics, NCJ-125833,
November 1990.
Reiber, Leslie, Parole in the United States 1980 and 1981, Washington: U.S. Bureau of
Justice Statistics, December 1985.
Report to the Nation on Crime and Justice, Washington: U.S. Bureau of Justice Statistics, NCJ-105506, March 1988.
Snell, Tracy L., Correctional Populations in the United States, 1992, NCJ-146413,
Washington: U.S. Bureau of Justice Statistics, January 1995.
Sourcebook of Criminal Justice Statistics, Washington: U.S. Government Printing Office, selected annual issues.
“State and County Probation: Systems in Crisis,” Washington: U.S. General Accounting Office, May 27, 1976.
Thibault, Edward A, and John J. Maceri, “N.Y.S. Proactive Probation Time Management,” Journal of Probation and Parole, 1986, pp. 17-22.
U.S. Bureau of the Census, Public Employment, Washington: U.S. Government Printing Office, selected issues.
U.S. Office of Management and Budget, Standard Industrial Classification Manual,
Washington: U.S. Government Printing Office, 1987.
U.S. President’s Commission on Law Enforcement and Administration of Justice, Task
Force Report: Corrections, Washington: U.S. Government Printing Office, 1967.
Bibliography—187
Juvenile Correctional
Institutions
Cahalan, Margaret Werner, Historical Correction Statistics in the United States, 18501984, Washington: U.S. Bureau of Justice Statistics, NCJ-102529, December 1986.
Children in Custody; Public Juvenile Facilities, 1987, Washington: U.S. Office of Juvenile Justice and Delinquency Prevention, October 1989.
Children in Custody, 1973, No. SD-JD-2F, Washington: U.S. Department of Justice,
December 1977.
Children in Custody—1975, No. SD-JD-4F, Washington: U.S. Law Enforcement Assistance Administration, December 1979.
Children in Custody, 1975-85, NCJ-114065, Washington: U.S. Bureau of Justice Statistics, May 1989.
Employment and Earnings, Washington: U.S. Bureau of Labor Statistics, January 1993.
Greenfeld, Lawrence A., Prisoners in 1989, Washington: U.S. Bureau of Justice Statistics, NCJ-122716, May 1990.
Innes, Christopher A., Population Density in State Prisons, Washington: U.S. Bureau
of Justice Statistics, Special Report, NCJ-103204, December 1986.
Irwin, John, and James Austin, Its About Time, San Francisco: National Council on
Crime and Delinquency, 1987.
“Juvenile Detention Using Staff Supervision Rather than Architectural Barriers,” Washington: U.S. General Accounting Office, December 1985.
“Juveniles Processed in Criminal Court and Case Dispositions,” GGD-95-170, Washington: U.S. General Accounting Office, August 1995.
Justice Expenditure and Employment: 1988, NCJ-125619, Bureau of Justice Statistics,
Washington: U.S. Government Printing Office, 1991.
Maguire, Kathleen and Ann L. Pastore, editors, Sourcebook of Criminal Justice Statistics 1993. Bureau of Justice Statistics, Washington: U.S. Government Printing Office,
1994.
Mancini, Norma, Our Crowded Jails: A National Plight, Washington: U.S. Bureau of
Justice Statistics, NCJ-111846, June 1988.
U.S. Bureau of the Census, Public Employment in 1971, Series GE-No. 1, Washington:
U.S. Government Printing Office, 1972.
U.S. Bureau of the Census, Public Employment: 1991, Series GE/91-1, Washington:
U.S. Government Printing Office, 1992.
U.S. Bureau of Justice Statistics, Report to the Nation on Crime & Justice, Washington:
U.S. Government Printing Office, March 1988.
U.S. General Accounting Office, “Juvenile Detention Using Staff Supervision Rather
than Architectural Barriers,” Washington: U.S. General Accounting Office, December
1985.
Bibliography—188
U.S. General Accounting Office, “Juveniles Processed in Criminal Court and Case Disposition,” GGD-95-170, Washington: U.S. General Accounting Office, August 1995.
U.S. Office of Management and Budget, Standard Industrial Classification Manual,
Washington: U.S. Government Printing Office, 1987.
Summary and Conclusions
Census of Local Jails-1988, Washington: Bureau of Justice Statistics, February 1990.
State Employment
Security Services
Unemployment Insurance
U.S. Bureau of the Census, Public Employment: 1992, Series GE/92-1, Washington:
U.S. Government Printing Office, 1994,
“Basic Structure of a Federal-State Unemployment Insurance Program and Related
Supporting Provisions,” Arlington, Virginia: National Commission on Unemployment
Compensation, November 1979.
Blaustein, Saul J., Unemployment Insurance in the United States, The First Half Century, Kalamazoo, Michigan: W.E. Upjohn Institute for Employment Research, 1993.
Burgess, Paul L. and Jerry L. Kingston, An Incentives Approach to Improving the Unemployment Compensation System, Kalamazoo, Michigan: W.E. Upjohn Institute for
Employment Research, 1987.
Burgess, Paul L. and Stuart A. Low, Unemployment Insurance and Employer Layoffs,
Unemployment Occasional Paper 93-1, Washington: U.S. Department of Labor, 1993.
“Clarification of UI Budget Workload Definitions,” Washington: U.S. Unemployment
Insurance Service, February 1, 1980.
“Development and Utilization of the Cost Model Management System in the Unemployment Insurance Program,” Washington: U.S. Employment and Training Administration, May 1977.
Diefenbach, Donald L., Financing America’s Unemployment Compensation Program,
Washington: U.S. Government Printing Office, 1979.
Dunson, Bruce H., S. Charles Maurice and Gerald P. Dwyer, Jr., The Cyclical Effects of
the Unemployment Insurance (UI) Program, Unemployment Occasional Paper 91-3,
Washington: U.S. Department of Labor, 1991.
Fisk, Donald M., “Modest Productivity Gains in State Unemployment Insurance Service,” Monthly Labor Review, January 1983, pp. 24-27.
Holen, Arlene and Stanley A. Horowitz, “The Effect of Unemployment Insurance and
Eligibility Enforcement on Unemployment,” Arlington, Virginia: Public Research Institute, April 1974.
“Millions Can be Saved by Improving the Productivity of State and Local Governments
Administering Federal Income Maintenance Assistance Programs,” Washington: U.S.
General Accounting Office, June 5, 1981.
National Commission on Unemployment Compensation, Unemployment Compensation: Final Report, Washington: U.S. Government Printing Office, 1980.
Bibliography—189
“Report on the Analysis of Initial Claims and Wage Record Activities by the Operational Improvement and Cost Equalization Project,” Washington: U.S. Unemployment
Insurance Service, 1979.
The Budget for Fiscal Year 1994, Appendix A, Washington: U.S. Government Printing
Office, 1994.
The Budget for Fiscal Year 1995, Appendix A, Washington: U.S. Government Printing
Office, 1995.
“SESA Accounting System Accounting Manual - Report Utilization Guide,” Washington: U.S. Employment and Training Administration, 1978.
“Unemployment Compensation: Studies and Research,” 3 volumes, Washington: U.S.
National Commission on Unemployment Compensation, July 1980.
“Unemployment Insurance: Administrative Funding is a Growing Problem for State
Programs,” GAO/HRD-89-72BR, Washington: U.S. General Accounting Office, May
1989.
“Unemployment Insurance—Inequities and Work Disincentives,” Washington: U.S.
General Accounting Office, 1979.
“Unemployment Insurance—Need to Reduce Unequal Treatment of Claimants and
Improve Benefit Payment Controls and Tax Collections,” Washington: U.S. General
Accounting Office, 1978.
“Unemployment Insurance: Opportunities to Strengthen the Tax Collection Process,”
GAO/HRD-89-5, Washington: U.S. General Accounting Office, June 1989.
Unemployment Insurance Quality Appraisal Results, an annual publication of the U.S.
Department of Labor, Employment and Training Administration.
“Unemployment Insurance: Trust Fund Reserves Inadequate,” GAO/HRD-88-55, Washington: U.S. General Accounting Office, September 1988.
“Unemployment Insurance: Trust Fund Reserves Inadequate to Meet Recession Needs,”
GAO/HRD-90-124, Washington: U.S. General Accounting Office, May 1990.
U.S. Bureau of Labor Statistics, Measuring Productivity in State and Local Government, Bulletin 2166, Washington: U.S. Government Printing Office, December 1983.
U.S. Bureau of Labor Statistics, Productivity Measures for Selected Industries and
Government Services, Bulletin 2440, Washington: U.S. Government Printing Office,
March 1994.
U.S. Manpower Administration, Handbook of Unemployment Insurance Financial
Data—1938-1970, Washington: U.S. Department of Labor, 1971.
U.S. Office of Management and Budget, Standard Industrial Classification Manual,
1987, Washington: U.S. Government Printing Office, 1987.
U.S. Unemployment Insurance Service, Twenty Years of Unemployment Insurance in
the USA—1935-1955, Washington: U.S. Government Printing Office, 1955.
Bibliography—190
Zajac, Wayne D. and David E. Balducchi (editors), Papers and Materials Presented at
the Unemployment Insurance Expert System Colloquium, June 1991, Unemployment
Occasional Paper 92-5, Washington: U.S. Department of Labor, 1992.
Employment Service
Adams, Leonard P., The Public Employment Service in Transition, 1933-1968, Ithaca,
New York: Cornell University, 1969.
“Allegations About the Ohio Bureau of Employment Service’s Operations in Cleveland, Ohio,” Washington: U.S. General Accounting Office, August 27, 1980.
Bain, Trevor, “Labor Market Analysis: A Review and Analysis of Manpower Research
and Development,” New York: Center for Policy Research, Inc., June 1975.
Benus, Jacob, and others, “Use of an Experimental Design in Assessing the Impact of
the United States Employment Service,” Menlo Park, California: Stanford Research
Institute, September 1976.
Cavin, Edward E. and Frank P. Stafford, “Efficient Provision of Employment Service
Outputs: A Production Frontier Analysis,” The Journal of Human Resources, Vol. XX,
No. 4, 1985, pp. 484-503.
Chadwin, Mark L., and others, The Employment Service: An Institutional Analysis,
Washington: U.S. Department of Labor, Employment and Training Administration, 1977.
Clark, William, “Production Costs and Output Qualities in Public and Private Employment Agencies,” The Journal of Law and Economics, October 1988, pp. 379-93.
Employment and Training Report of the President, selected issues, Washington: U.S.
Government Printing Office.
“Employment Service—Improved Leadership Needed for Better Performance,” Washington: U.S. General Accounting Office, August 1991.
“Employment Service—Leadership Needed to Improve Performance,” Washington: U.S.
General Accounting Office, October 1990.
“Employment Service—Variations in Local Office Performance,” Washington: U.S.
General Accounting Office, August 1989.
“Employment Service Performance Handbook for Local Offices,” Rockville, Maryland: WESTAT, Inc., 1979.
ESARS Handbook, Washington: U.S. Employment and Training Administration, November 1979, revision.
Fairchild, Charles K., A Performance Needs Based Methodology for Allocating Employment Service Grants, Cambridge, Massachusetts: Abt Associates, Inc., April 1980.
Frey, Donald E., A Methodology for Measuring the Impact of the United States Employment Service, Winston Salem, N.C.: Wake Forest University, 1976.
“Glossary of Program Terms and Definitions,” Washington: U.S. Employment and Training Administration, 1978.
Guzda, Henry P., “The U.S. Employment Service at 50: It too Had to Wait Its Turn,”
Monthly Labor Review, June 1983, pp. 12-19.
Bibliography—191
Historical Statistics of Employment Security Activities—1938-66, Washington: U.S.
Bureau of Employment Security, 1968.
Jones, Melvin and Toikka, Richard S., “A Review of Previous Research on the Determinants of Employment Service Productivity and Proposed Extensions,” Washington:
The Urban Institute, May 5, 1978.
Katz, Arnold, Explanatory Measures of Labor Market Influences of the Employment
Service, Pittsburgh: University of Pittsburgh, 1978.
Meike, C., and others, SESA Productivity Measurement System, Vienna, Virginia: Analytic Systems, September 24, 1976.
Moore, Basil, “A Benefit-Cost Analysis of the United States Employment Service,”
Washington: U.S. Bureau of the Budget, November 1966.
Stevens, David W. and others, “Specification and Measurement of Productivity in the
USES,” Washington: U.S. Employment Service, December 1980.
“Summary of Employment Security Statistical Reports, Washington: U.S. Employment
and Training Administration, August 1977.
“The Employment Service—Problems and Opportunities for Improvement,” Washington: U.S. General Accounting Office, 1977.
The Impact of Employment Service Regulations on the ES Network, Phase I Findings,
Booz Allen & Hamilton, Inc., 1980.
Thorpe, Charles O., and Richard W. Toikka, Determinants of State Employment Service Productivity, Washington: The Urban Institute, March 1979.
U.S. Bureau of Labor Statistics, Measuring Productivity in State and Local Government, Bulletin 2166, Washington: U.S. Government Printing Office, December 1983.
U.S. Congress, House of Representatives, Committee on Government Operations, “Operation of the U.S. Employment Service,” Washington: U.S. Government Printing Office, 1976.
U.S. Congress, House of Representatives, Committee on Ways and Means, Subcommittee on Human Resources, “Reforming the Unemployment Compensation System,”
Washington: U.S. Government Printing Office, 1990.
U.S. Senate, Subcommittee of the Committee on Appropriations, “Labor-Health, Education, Welfare Appropriation for Fiscal Year 1968” (90th Congress, 1st session), Washington: U.S. Government Printing Office, 1967.
U.S. Office of Management and Budget, Standard Industrial Classification Manual
1987, Washington: U.S. Government Printing Office, 1987.
Weiner, Neil S., John H. Powel, and C. Michael Tahm, The United States Employment
Service: A Conceptual Model of Outputs, Values and Illustrative Estimations, Vols., I
and II, Arlington, Virginia: Boeing Computer Services, Inc., December, 1976.
Bibliography—192
Appendix A. Comparison of Bureau of the Census
Classification of State and Local Government
Functions with Standard Industrial Classification
Government function
(Bureau of the Census)
Description
SIC industry
Air transportation
Operation and support of publicly operated
airport facilities.
4581 Airports, flying fields, and airport
terminal services
Correction
Activities pertaining to the confinement and
correction of adults and minors convicted
of criminal offenses. Pardon, probation,
and parole activities are also included here.
8322 Individual and family social
services
8361 Residential care
9223 Correctional institutions
Electric power
Activities associated with the production or
acquisition and distribution of electric power.
4911
Elementary and secondary
education
All activities associated with the operation of
public elementary and secondary schools and
locally operated vocational-technical schools.
Special education programs operated by
elementary and secondary school systems
are also included, as are all ancillary services
associated with the operation of schools,
such as pupil transportation and food service.
4151 School buses
8211 Elementary and secondary schools
9411 Administration of educational
programs
Financial administration
Includes activities concerned with tax
assessment and collection, custody and
distribution of funds, debt management,
administration of trust funds, budgeting,
and other government wide financial
management activities. This function is not
applied to school district or special district
governments.
9311
Fire protection
Applies to local government fire protection
and prevention activities plus any ambulance,
rescue, or other auxiliary services provided by
the fire protection agency.
9224 Fire protection
Gas supply
Local government activities associated with
the acquisition of gas supplies and
distribution to individual consumers.
4924 Natural gas distribution
Health
Administration of public health programs,
community and visiting nurse services, immunization programs, drug abuse rehabilitation programs, health and food inspection
activities, operation of outpatient clinics,
and environmental pollution control
activities.
8082 Home health care services
8093 Specialty outpatient facilities,
nec
9431 Administration of public health
programs
9641 Regulation of agricultural marketing and commodities
Electric services
Public finance, taxation and
monetary policy
Appendix A—193
Appendix A. Comparison of Bureau of the Census Classification of State and Local Government
Functions with Standard Industrial Classification—Continued
Government function
(Bureau of the Census)
Description
SIC industry
Higher education
Includes local government degree-granting
institutions which provide academic training
above grade 12.
8221 Colleges, universities and
professional schools
8222 Junior colleges and technical
institutes
Highways
Activities associated with the maintenance
and operation of streets, roads, sidewalks,
bridges, tunnels, and toll roads, and ferries.
Snow and ice removal, street lighting,
and highway and traffic engineering
activities are also included here.
1611 Highways and street construction
1622 Bridge, tunnel, and elevated
highway construction
4482 Ferries
4785 Fixed facilities and inspection
and weighing services
Hospitals
Includes only government-operated medical
care facilities which provide inpatient care.
8062 General medical and surgical
hospitals
8063 Psychiatric hospitals
8069 Specialty hospitals, except
psychiatric
Housing and community
development
The operation of housing and redevelopment
projects and other activities to promote or aid
housing and community development.
6513 Operators of apartment buildings
9531 Administration of housing
programs
9532 Administration of urban planning
and community and rural development
Judicial and legal
Includes all court and court related activities
(except probation and parole activities which
are included at the “Correction” function),
court activities of sheriff’s offices, prosecuting
attorneys’ and public defender’s officers, legal
departments, and attorneys providing
government legal service.
9211 Courts
9222 Legal counsel and prosecution
9229 Public order and safety, nec
Libraries
Applies only to libraries operated by local
governments for use by the general public.
School and law libraries are included in
“Elementary and secondary education” or
“Higher education” and “Judicial and legal”
categories respectively.
8231 Libraries
Natural resources
Activities primarily concerned with the
conservation and development of natural
resources—forest fire prevention and control,
flood control, irrigation, drainage, land and
forest reclamation, fish and game preservation
and control, soil conservation, forestry,
agricultural aids and research, agriculture
development and inspection, and mineral
resources.
0851 Forestry services
0921 Fish hatcheries and preserves
0971 Hunting and trapping, and game
propagation
4971 Irrigation systems
9512 Land, mineral, wildlife, and
forest conservation
9631 Regulation and administration of
utilities
9641 Regulation of agricultural
marketing and commodities
Appendix A—194
Appendix A. Comparison of Bureau of the Census Classification of State and Local Government
Functions with Standard Industrial Classification—Continued
Government function
(Bureau of the Census)
Description
SIC industry
Other education
State government activities relating to the
supervision and regulation of public and private
elementary and secondary schools; programs
and institutions for the training of blind, deaf,
and other handicapped persons; and vocational
rehabilitation programs.
8249 Vocational schools, nec
9411 Administration of educational
programs
Other government
administration
Applies to the legislative and governmentwide administrative agencies of government.
Included here are overall planning and zoning
activities, and central personnel and
administrative activities.
9111
9121
9131
9199
Parks and recreation
Government activities that include
the operation and maintenance of parks,
playgrounds, swimming pools, public beaches,
auditoriums, public golf courses, museums,
marinas, botanical gardens, and zoological
parks.
0782
4493
7992
7999
Lawn and garden services
Marinas
Public golf courses
Amusement and recreation
services, nec
8412 Museums and art galleries
8422 Arboreta and botanical or
zoological gardens
9512 Land, mineral, wildlife, and
forest conservation
Police protection
All activities concerned with the enforcement
of law and order, including coroners’ offices,
police training academies, investigation
bureaus, and local jails, “lockups,” or
other detention facilities not intended to
serve as correctional facilities.
9221 Police protection
9229 Public order and safety, nec
Public welfare
Included in this category are such activities
as the administration of various public
assistance programs for the needy, operation
of homes for the elderly and indigent, and
administration of programs that provide
payments for medical care and other services
for the needy. Health care and hospital
services provided directly by a government,
however, are included in the “Health” and
“Hospital” functions rather than here.
8322 Individual and family social
services
8351 Child day care services
8361 Residential care
8399 Social services, nec
9441 Administration of social, human
resource and income maintenance
programs
Sanitation other than
sewerage
Refuse collection and disposal, operation of
sanitary landfills, and street cleaning activities.
4212
4953
4959
9511
Sewerage
The provision, maintenance, and operation of
sanitary and storm sewer systems and sewage
disposal and treatment facilities.
4952 Sewerage systems
9511 Air and water resource and
solid waste management
Executive offices
Legislative bodies
Executive &legislative offices
General government, nec
Local trucking without storage
Refuse systems
Sanitary services, nec
Air and water resource
and solid waste management
9631 Regulation and administration
of utilities
Appendix A—195
Appendix A. Comparison of Bureau of the Census Classification of State and Local Government
Functions with Standard Industrial Classification—Continued
Government function
(Bureau of the Census)
Description
SIC industry
Social insurance
administration
The administration and conduct of social
insurance programs. For State governments,
these activities include unemployment
compensation, worker compensation,
and work study programs.
7361 Employment agencies
8331 Job training and vocational
rehabilitation services
9441 Administration of social,
human resource and income
maintenance programs
State liquor stores
The administration and operation of liquor
stores by State governments.
5182 Wine and distilled alcoholic
beverages(wholesale)
5921 Liquor stores
9651 Regulation, licensing and
investigation
Transit
Activities relating to the operation and
maintenance of public mass transit
systems (e.g., bus, subway, surface rail,
and street railroad systems).
4111 Local and suburban transit
4119 Local passenger transportation, nec
9621 Regulation and administration
of transportation programs
Water supply
Local government activities associated with
the production or acquisition of water and
distribution to the public.
4941 Water supply
Water transport and
terminals
Activities which are connected with the
operation and support of canals and
other waterways, harbors, docks,
wharves, and other related marine terminal
facilities.
4491 Marine cargo handling
NOTE: nec = not elsewhere classified
SOURCE: 1992 Census of Governments-Compendium of Public Employment (Bureau of the Census, 1997), appendix A, and Standard Industrial Classification Manual, 1987 (Office of Management and Budget, 1987).
Appendix A—196